magmatic processes during the prolonged pu'u 'o'o eruption of
TRANSCRIPT
JOURNAL OF PETROLOGY VOLUME 41 NUMBER 7 PAGES 967–990 2000
Magmatic Processes During the ProlongedPu’u ’O’o Eruption of Kilauea Volcano,Hawaii
MICHAEL O. GARCIA1∗, AARON J. PIETRUSZKA1†, J. M. RHODES2
AND KIERSTIN SWANSON1
1HAWAII CENTER FOR VOLCANOLOGY, DEPARTMENT OF GEOLOGY AND GEOPHYSICS, UNIVERSITY OF HAWAII,
HONOLULU, HI 96822, USA2DEPARTMENT OF GEOSCIENCES, UNIVERSITY OF MASSACHUSETTS, AMHERST, MA 01003, USA
RECEIVED SEPTEMBER 20, 1999; REVISED TYPESCRIPT ACCEPTED FEBRUARY 8, 2000
KEY WORDS: Kilauea Volcano; crystal fractionation; magma mixing;The Pu’u ’O’o eruption is exceptional among historical eruptionsmantle melting processesof Kilauea Volcano for its long duration (>17 years and continuing),
large volume (>2 km3), wide compositional range (5·6–10·1 wt% MgO) and the detailed monitoring of its activity. The prolongedperiod of vigorous effusion (>300 000 m3/day) and the simplephenocryst mineralogy of the lavas (essentially only olivine) has INTRODUCTIONallowed us to examine the volcano’s crustal and mantle magmatic The Pu’u ’O’o eruption is the longest-lived, best monitoredprocesses. Here we present new petrologic data for lavas erupted (e.g. Wolfe et al., 1988) and most voluminous historicalfrom 1992 to 1998 and a geochemical synthesis for the overall eruption of Kilauea Volcano (Fig. 1). It started in Januaryeruption. The dominant crustal magmatic processes are fractionation 1983 and, 17 years later, it is still continuing vigorouslyand accumulation of olivine, which caused short-term (days to (>300 000 m3/day). About>2 km3 of basaltic lava (denseweeks) compositional variations. Magma mixing was important rock equivalent) have been erupted, which destroyed 181only during the early part of the eruption and during episode 54. homes and a National Park visitor center on the south flankThe overall systematic decrease in MgO-normalized CaO content of the volcano (total losses >$60 million). We previouslyand abundances of highly incompatible elements, without significant documented the petrologic history of lavas from episodesPb, Sr and Nd isotope compositional variation, is interpreted to be 1 to 49 of this eruption (1983 to early 1992) utilizingcaused by mantle melting processes. Experimental results and petrography, mineral and whole-rock geochemistry [elec-modeling of trace element variations indicate that neither batch tron microprobe, X-ray fluorescence (XRF) and inductivelymelting nor simple progressive melting can explain these compositional coupled plasma mass spectrometry (ICP-MS) analyses],variations. Instead, a more complex progressive melting model is and Pb, Sr, Nd and O isotopes supplemented with fieldneeded. This model involves two source components with the same observations and geophysical data (Garcia & Wolfe, 1988;isotopic composition, but one was melted >3% in the Hawaiian Garcia et al., 1989, 1992, 1996, 1998a). This paper chron-plume. The model results indicate that the amount of this depleted icles the petrology and geochemistry (including ICP-MSsource component progressively increased during the eruption from 0 trace element data, and Pb, Sr and Nd isotopic ratios) ofto >25%. Given the isotopic similarity of Pu’u ’O’o lavas to Pu’u ’O’o lavas from episodes 50 to 55 (1992 to earlymany lavas from Loihi Volcano and the small extent of prior melting 1998), a period when >0·9 km3 of lava was produced. Into form the depleted source component, the melting region for Pu’u addition, we present the results of a time-series analysis of’O’o magmas may partially overlap with that of the adjacent, the entire eruption, which provides a rare opportunity to
document both crustal magmatic processes (e.g. crystalyounger volcano, Loihi.
∗Corresponding author. Telephone: +1-808-956-6641. Fax:+1-808-956-5512. e-mail: [email protected]†Present address: Department of Terrestrial Magnetism, CarnegieInstitution of Washington, Washington, DC 20015, USA.Extended data set can be found at:http://www.petrology.oupjournals.org Oxford University Press 2000
JOURNAL OF PETROLOGY VOLUME 41 NUMBER 7 JULY 2000
Fig. 1. Map of the Pu’u ’O’o eruption flow field on the east rift zone of Kilauea Volcano, Hawaii (after Heliker et al., 1998b). Lava has eruptedprimarily from two central vents: Pu’u ’O’o (episodes 2–47, 50–53 and 55) and Kupaianaha (episode 48). A fissure system formed between thesevents during episode 49 and uprift from the Pu’u ’O’o vent during episode 54. The lavas from the continuing episode 55 are not shown. Thecontour interval is 500 ft after the 100 ft contour (1 ft = 0·3048 m). The insert map shows the location of the eruption site on the island ofHawaii.
fractionation, magma mixing and crustal assimilation) and the source components, the southward dip of Kilauea’smantle conduit, and the indications from U-series datato investigate mantle melting systematics. In contrast to thethat Kilauea magmas are being derived from a largebrief, small-volume, historical eruptions (1790–1982; seesource area are all consistent with the possibility thatMacdonald et al., 1983), long-lived eruptions such as Pu’uthese magmas were derived from the same source region’O’o may have typified Kilauea’s prehistoric history (e.g.that is supplying the adjacent, younger volcano, Loihi.Holcomb, 1987).
Our results show that olivine fractionation and ac-cumulation are the dominant crustal processes duringthe Pu’u ’O’o eruption, although crustal assimilation and
BRIEF HISTORY OF THE PU’U ’O’Omagma mixing have played important secondary rolesERUPTIONduring the early part of the eruption and during episode
54. A progressive change in whole-rock CaO con- The character of the Pu’u ’O’o eruption has changedcentration and in incompatible trace element abundances markedly during 17 years of activity (see Table 1 for aand ratios with essentially no change in Pb, Sr and Nd summary). Episode 1 involved a ‘curtain of fire’ along aisotopic ratios indicates that mantle partial melting was discontinuous fissure system of 8 km length, which wasthe primary factor controlling compositional variation in intermittently active for almost a month (Wolfe et al.,Pu’u ’O’o lavas erupted from 1985 to 1998. Modeling 1987). Episodes 2–47 (February 1983–June 1986) wereof this compositional variation requires melting of an short lived (5–100 h) with lava fountains of variableisotopically homogeneous source with a component that heights (10–400 m). These episodes were followed bywas previously melted >3%. The model results suggest repose periods of 8–65 days during which magma ac-that the proportion of this component in the lavas in- cumulated and fractionated in a shallow reservoir undercreased from 0 to>25% during the Pu’u ’O’o eruption. the Pu’u ’O’o vent (Wolfe et al., 1988; Hoffmann et al.,The isotopic similarity of Pu’u ’O’o lavas to many Loihi 1990; Garcia et al., 1992). In July 1986, the site of effusive
activity shifted 3 km down-rift from the 255-m-high Pu’ulavas, the small extent of previous melting of one of
968
GARCIA et al. MAGMATIC PROCESSES FOR A PROLONGED KILAUEA ERUPTION
Table 1: Summary of the Pu’u ’O’o eruption, 1983–1999
Episode Episode Repose Episode Volume Vent location Eruption rate
start date length (days) length (106 m3) (103 m3/day)
1 Jan. 03, 1983 start 20 days 14 Fissure 1 —
2–47 Feb. 10, 1983 8–65 >3·5 years 371 Mostly Pu’u ’O’o >300
48 July 20, 1986 24 >5·5 years >500 Kupaianaha >400–0·5
49 Nov. 08, 1987 none 18 days 11 Fissure 2 0·6
50 Feb. 17, 1992 11 15 days 3 Pu’u ’O’o flank —
51 Mar. 07, 1992 4 161 days 32 Pu’u ’O’o flank >300
52 Oct. 03, 1992 none 15 days 2 Pu’u ’O’o flank >300
53 Feb. 20, 1993 none >4 years >535 Pu’u ’O’o flank >300
54 Jan. 30, 1997 none >1 day 0·3 Fissure 3 0·3
55 Feb. 24, 1997 24 continuing >265∗ Pu’u ’O’o and its 200–500
flank
Repose length refers to the duration of the pause between eruptive episodes; volume is the dense rock equivalent eruptedduring each episode. Vent locations (see Fig. 1): fissure 1—a discontinuous fissure system of 8 km length extending eastfrom Napau crater under the Pu’u ’O’o cone; fissure 2—a discontinuous fissure system of 2 km length between the Pu’u’O’o and Kupaianaha vents; Pu’u ’O’o flank—lava rose up in the throat of Pu’u ’O’o vent and drained onto the flow fieldthrough vents on the south flank of Pu’u ’O’o for episodes 52 and 55 and west flank for episodes 50, 51, 53 and 55; fissure3—a discontinuous fissure of 2 km length extending from the west wall of Napau Crater eastward towards Pu’u ’O’o. Datasources: Episodes 1–20, Wolfe et al. (1988); episodes 21–47, Ulrich et al. (1987); episode 48, Kauahikaua et al. (1996); episode49, Mangan et al. (1995); episodes 50–53, Heliker et al. (1998b); episode 54, Harris et al. (1997); episode 55, Heliker et al.(1998a). The eruption rate values are longer-term averages for each episode except episode 48, which showed a decreasefrom 250 000 m3/day to 54 000 m3/day during the 8 months before the Kupaianaha vent died; the values for episodes 51–53are an average for only the period Nov. 1992–Feb. 1994.∗Estimated value as of January 1999.
’O’o cone to form the Kupaianaha vent (Fig. 1). Magma near Pu’u ’O’o (Heliker et al., 1998b). The next day,episode 52 began from two new vents on the southwestrose up in the throat of the Pu’u ’O’o cone and degassedflank of the cone, >200–300 m from the episode 51vigorously before passing through a shallow undergroundvent (which resumed erupting the next day). Episode 52channel to the Kupaianaha vent (Garcia et al., 1996).continued until a seismic swarm and a rapid summitThis shift in the vent site marked a fundamental changedeflation in February 1993 (Heliker et al., 1998b). Episodein eruption style from episodic, high fountaining to nearly53 started 9 days later and continued for almost 4 yearscontinuous, quiescent effusion. The Kupaianaha vent wasuntil the Pu’u ’O’o lava lake suddenly drained and theactive for 5·5 years during which it produced>0·5 km3 ofcone collapsed on January 29, 1997. A few hours later,lava (dense rock equivalent) and built a shield of 56 ma 1 day eruption (episode 54) occurred along a dis-height during this 48th episode of the Pu’u ’O’o eruptioncontinuous fissure system of 2 km length, 2–4 km up-rift(Kauahikaua et al., 1996). Three months before the endfrom Pu’u ’O’o. Lava was produced during episode 54of episode 48, a discontinuous fissure system of 1·7 kmfrom six, low fountaining (10–30 m high) vents (Harris etlength opened up between the Pu’u ’O’o and Kupaianahaal., 1997). A 24 day hiatus followed episode 54. Duringvents (episode 49). For approximately 21 days, this ventthis period, no lava was observed in or near the Pu’usystem produced lavas of the same composition as coeval’O’o cone and many volcano watchers thought the erup-lavas from the Kupaianaha vent (Garcia et al., 1996).tion was finally over. This eruption, however, is con-Activity at the Kupaianaha vent diminished during epi-tinuing (episode 55) and shows no signs of ending in thesode 49 and ended 2 months later.near future despite occasional short (1–4 day) eruptiveWith the demise of the Kupaianaha vent in early 1992,pauses.effusive activity resumed 11 days later at Pu’u ’O’o from
vents on its flanks. Episode 50 was short lived (15 days)but the eruption restarted (episode 51) along the same
SAMPLESfissure after a 4 day hiatus (Table 1). Episode 51 wasinterrupted by 17 short pauses (8–90 h) and ended Ninety-three lava samples representing episodes 50–53,
six from episode 54 and an additional 29 from episode>7 months later following a magnitude 4·5 earthquake
969
JOURNAL OF PETROLOGY VOLUME 41 NUMBER 7 JULY 2000
55 were collected for this study. Whenever possible, rare (Ζ0·2 vol. %) phenocrysts of olivine and plagioclase.Rare microphenocrysts of plagioclase, olivine and clino-samples were collected in a molten state from active lava
flows and water-quenched to minimize post-eruptive pyroxene are also present in these lavas.Xenoliths are rarely reported in Kilauea lavas but arecrystallization. The episode 54 samples were collected
several weeks after their eruption, but from rapidly present in six of the samples we studied in thin section.These small basalt xenoliths contain ‘black’ olivine. Blackquenched vent deposits and are labeled according to
their eruptive vent (A–F). All other samples are labeled olivines are created by high-temperature oxidation asforsteritic olivine (>90%), magnetite and hypersthenewith the date that they were collected (day-month-year),
which is probably within a day of their eruption. During replace the original olivine (Macdonald, 1944).episodes 50–52, lava samples were mostly collected within100 m of the eruptive vent. Following the onset of episode53, lava drained directly into tubes and only rarely were
MINERAL CHEMISTRYoverflows and skylights available for sampling. Therefore,Olivine and clinopyroxene compositions were determinedmost of our samples from episode 53 and all from episodefor a suite of episode 50–55 lavas that span nearly the55 were collected on the coastal plain, 10–12 km fromentire range of whole-rock MgO contents (6·1–10·1 wtthe Pu’u ’O’o vent, as the lava discharged from lava%). Plagioclase compositions were determined for thetubes. Splits of our Pu’u ’O’o samples are available totwo episode 54 lavas with plagioclase phenocrysts. Theanyone interested in performing additional analyses.University of Hawaii, five-spectrometer, Cameca SX-50electron microprobe with SAMx automation was usedfor the mineral analyses. Operating conditions were15 kV, a minimum spot size of 1 �m and a beam currentPETROGRAPHYof 20 nA for olivine and pyroxene, and 10 nA forEpisode 50–53 and 55 lavas, like the vast majority ofplagioclase. Each element was counted for 30–45 s onother Pu’u ’O’o lavas we have studied, are glassy, stronglythe peak and 15 s on the background. Concentrationsvesicular, friable and weakly to moderately olivine-phyricwere determined using a ZAF–PAP correction scheme.(0·4–6·2 vol. % phenocrysts; Table 2). Olivine is theAnalytical error is estimated to be>0·2% forsterite, andonly phenocryst in these samples; it is usually small1–2% relative for major elements and 5–10% relative(>0·1–1 mm in diameter), euhedral, undeformed andfor minor elements (<1 wt %) based on repeated analysescontains spinel and glass inclusions. Olivine is somewhatof standards. Three spot analyses were made in themore common (>1 vol. %) in these lavas compared withcore and one on the rim of each crystal to check forthose from the preceding 5·5 years of eruptive activitycompositional zoning.at the Kupaianaha vent (episode 48), despite identical
ranges in MgO contents (7–10 wt %; Garcia et al., 1996)for the lavas from the two periods. Olivine abundance
Olivinein episodes 50–53 and 55 lavas is generally correlatedpositively with whole-rock mg-number, although olivine Over 200 olivine crystals were analyzed from 19 episodephenocryst abundance can vary as much as 5 vol. % for 50–55 lavas. Most of the analyzed olivine crystals area given mg-number (Table 2). normally zoned with <3% variation in forsterite (Fo)
Clinopyroxene (cpx) microphenocrysts occur in>65% from core to rim. The forsterite content of the olivineof the episode 50–53 lavas and in many of the early cores from the episode 50–53 and 55 lavas range fromepisode 55 lavas that we examined but always in low 79·7 to 84·4% (Table 3) with a few notable exceptionsabundance (<1·7 vol. %). Commonly, these small (e.g. black olivines with Fo >90%). Phenocrysts and(0·1–0·3 mm) crystals occur in clusters of 3–12 grains and microphenocrysts in episode 50–53 lavas overlap in com-some have sector zoning. Plagioclase microphenocrysts position (average Fo contents of >81·0%) and have(0·1–0·5 mm) are absent in the episode 50–52 samples somewhat lower Fo contents than the olivines fromwe examined but rare (Ζ0·4 vol. %) crystals occur in episode 48 lavas (>82·5% average; Garcia et al., 1996).>33% of the episode 53 samples. Tiny crystals (<0·1 mm) Olivines from episode 55 lavas have somewhat higherof plagioclase, olivine, spinel and clinopyroxene occur in forsterite contents then episode 50–53 olivine (averagea matrix of honey brown glass or black cryptocrystalline 82·4%) and have a compositional distinction betweenmaterial in nearly all samples. Episode 55 samples col- phenocrysts and microphenocrysts (average 83·5 vslected after April 21, 1997, when surface activity became 81·7% Fo, respectively).more vigorous, do not contain cpx or plagioclase micro- Cores of olivines from lavas with a wide range of mg-phenocrysts. numbers (56–61) have olivine compositions in equilibrium
The episode 54 lavas are among the most aphyric with the bulk rock (Fig. 2). Many episode 50–53 and 55lavas, especially those with higher mg-numbers, containsamples from the Pu’u ’O’o eruption. They contain only
970
GARCIA et al. MAGMATIC PROCESSES FOR A PROLONGED KILAUEA ERUPTION
Table 2: Modal mineralogy of representative lavas from episodes 50 to 55 of the Pu’u ’O’o eruption
Sample mg-no. Olivine Clinopyroxene Plagioclase Matrix
ph mph mph ph mph
Episode 50
19-Feb-92 61·1 2·4 3·4 <0·1 0·0 0·0 94·2
23-Feb-92 61·8 0·4 2·4 <0·1 0·0 0·0 97·2
Episode 51
13-Jun-92 61·2 2·0 3·8 1·0 0·0 0·0 93·2
05-Oct-92 58·6 2·4 1·2 0·4 0·0 0·0 96·0
29-Dec-92∗ 59·7 2·0 2·8 0·2 0·0 0·0 95·0
Episode 52
04-Oct-92 60·8 1·6 3·2 1·6 0·0 0·0 93·6
Episode 53
18-Mar-93∗ 60·2 2·2 3·2 0·2 0·0 0·0 94·4
16-Apr-93 60·1 1·2 2·2 0·0 0·0 0·0 96·6
09-May-93 61·1 1·8 2·8 0·0 0·0 0·0 95·4
30-Oct-93 56·5 0·6 1·0 0·2 0·0 0·4 97·8
14-Dec-93 60·2 0·8 2·0 1·4 0·0 0·0 95·8
13-Mar-94 59·1 0·8 2·2 0·4 0·0 0·0 96·6
18-May-94 57·5 2·4 3·2 1·6 0·0 <0·1 92·8
09-Oct-94 59·0 2·0 3·0 0·6 0·0 0·0 94·4
14-Jan-95∗ 59·0 2·4 3·0 1·4 0·0 0·0 93·2
29-Mar-95 58·4 1·6 2·2 0·4 0·0 0·0 95·8
14-Jun-95 61·3 3·6 2·6 0·6 0·0 0·0 93·2
11-Sep-95∗ 56·1 0·8 2·0 0·2 0·0 0·0 97·0
14-Oct-95∗ 60·5 2·6 3·8 0·4 0·0 0·4 92·8
19-Jan-96 63·4 6·2 2·4 0·6 0·0 0·2 90·6
15-Mar-96 59·2 2·8 5·0 0·2 0·0 <0·1 92·0
07-Jun-96 58·0 2·6 1·8 <0·1 0·0 <0·1 95·6
22-Aug-96 59·2 2·6 2·6 0·0 0·0 0·0 94·8
10-Jan-97 59·5 2·2 3·0 0·0 0·0 0·0 94·8
Episode 54
B vent 51·7 <0·1 <0·1 0·2 0·0 0·4 99·4
D vent 52·4 <0·1 <0·1 <0·1 0·2 0·6 99·2
F vent 51·9 0·0 <0·1 0·4 0·0 1·6 98·0
Episode 55
21-Apr-97 56·7 0·6 1·8 0·6 0·0 0·2 96·8
10-Aug-97∗ 62·9 5·0 5·4 0·0 0·0 0·0 89·6
11-Oct-97 58·0 0·8 1·0 0·0 0·0 0·0 98·2
26-Dec-97 61·6 2·6 3·4 0·0 0·0 0·0 94·0
10-Jan-98 61·3 1·8 4·4 0·0 0·0 0·0 93·8
27-Mar-98 60·6 3·2 2·2 0·0 0·0 0·0 94·6
10-Apr-98 61·3 1·4 4·2 0·0 0·0 0·0 94·4
11-May-98 60·8 5·2 3·0 0·0 0·0 0·0 91·8
17-Aug-98 60·8 1·4 3·4 0·0 0·0 0·0 95·2
mg-no. is [Mg/(Mg + Fe2+) × 100], assuming Fe2+ = 0·9 of total iron. All values are in vol. % and are based on 500 pointcounts/sample. Phenocrysts (ph) are >0·5 mm across; microphenocrysts (mph) are 0·1–0·5 mm across. Matrix consists ofglass and crystals <0·1 mm across.∗Contains black olivine in basalt xenolith(s).
971
JOURNAL OF PETROLOGY VOLUME 41 NUMBER 7 JULY 2000
Tab
le3
:R
epre
sent
ativ
em
icro
prob
ean
alys
esof
oliv
ines
from
epis
ode
50–
55
lava
sof
the
Pu’
u’O
’oer
uption
Ep
iso
de:
5051
5152
Sam
ple
:23
-Feb
-92
29-D
ec-9
213
-Ju
n-9
204
-Oct
-92
Ro
ckm
g-n
o.:
61·8
59·7
61·2
60·8
Siz
e:m
ph
mp
hm
ph
ph
-cm
ph
ph
-cp
h-c
mp
hp
h-c
mp
h
SiO
239
·55
39·3
539
·15
39·4
38·7
539
·25
39·2
39·0
39·4
539
·15
FeO
15·9
517
·017
·617
·018
·117
·418
·118
·216
·55
17·5
0N
iO0·
110·
180·
150·
200·
170·
230·
200·
190·
220·
21M
gO
43·7
543
·45
42·8
043
·60
43·0
42·8
42·5
542
·65
43·9
542
·70
CaO
0·29
0·27
0·30
0·27
0·31
0·27
0·27
0·27
0·26
0·24
Tota
l99
·65
100·
2510
0·00
100·
4710
0·3
99·9
510
0·32
100·
3110
0·43
99·8
Fom
ol%
83·0
82·0
81·3
82·1
81·0
81·4
80·7
80·7
82·5
81·3
Ep
iso
de:
5353
5353
Sam
ple
:18
-Mar
-93
09-M
ay-9
313
-Mar
-94
29-M
ar-9
5R
ock
mg
-no
.:60
·261
·159
·158
·4
Siz
e:b
kp
h-c
bk
mp
hp
h-c
ph
-rm
ph
ph
-cp
h-r
mp
hp
h-c
ph
-rm
ph
SiO
242
·842
·639
·25
39·1
538
·539
·25
39·3
39·3
39·4
38·4
39·2
FeO
2·4
2·85
16·4
517
·717
·55
17·8
518
·018
·017
·419
·95
18·5
NiO
0·21
0·12
0·24
0·17
0·17
0·17
0·15
0·18
0·22
0·15
0·18
Mg
O54
·95
54·1
43·9
543
·143
·042
·242
·342
·343
·040
·642
·2C
aO0·
130·
210·
250·
280·
300·
280·
270·
270·
260·
320·
26To
tal
100·
4999
·88
100·
1410
0·4
99·5
299
·75
100·
0210
0·05
100·
2899
·42
100·
34Fo
mo
l%97
·496
·982
·680
·581
·380
·880
·780
·781
·578
·480
·3
Ep
iso
de:
5454
5555
Sam
ple
:B
ven
tD
ven
t10
-Au
g-9
711
-May
-98
Ro
ckm
g-n
o.:
51·7
52·4
62·9
60·8
Siz
e:m
ph
mp
hm
ph
mp
hb
km
ph
ph
-cxe
no
ph
-cp
h-c
mp
h
SiO
239
·75
40·2
40·3
38·9
41·5
39·3
38·9
39·6
41·0
39·8
FeO
16·1
13·9
13·7
19·3
4·7
16·8
521
·75
16·0
10·3
16·9
NiO
0·20
0·26
0·23
0·01
0·19
0·18
0·14
0·26
0·41
0·23
Mg
O43
·044
·65
44·8
540
·153
·042
·839
·243
·748
·542
·3C
aO0·
250·
250·
250·
270·
150·
260·
250·
250·
250·
48To
tal
99·3
99·2
699
·33
99·5
899
·55
99·4
410
0·24
99·8
510
0·45
99·7
1Fo
mo
l%82
·685
·185
·378
·890
·281
·676
·082
·789
·181
·2
Ro
ckm
g-n
o.
isw
ho
le-r
ock
[Mg
/(Fe
2++
Mg
)]×
100
assu
min
g90
%o
fth
eto
tal
iro
nis
Fe2+
;p
h,
ph
eno
crys
ts;
mp
h,
mic
rop
hen
ocr
ysts
;c,
core
;r,
rim
;b
k,b
lack
oliv
ine;
xen
o,
xen
ocr
yst.
All
oxi
de
con
cen
trat
ion
sar
ein
wt
%.
972
GARCIA et al. MAGMATIC PROCESSES FOR A PROLONGED KILAUEA ERUPTION
Fig. 2. Whole-rock mg-number [(Mg/(Mg + Fe2+) × 100] vs olivine composition in forsterite content (Fo %) for crystal cores from episode50–55 lavas. For the mg-number calculation, 90% of the total iron is assumed to be Fe2+ based on iron redox determinations of Hawaiiantholeiites (Moore & Ault, 1965). Β, phenocrysts (0·5–2 mm in diameter); •, microphenocrysts (0·1–0·5 mm). The date below each vertical setof olivine data is the sample number. The diagonal stippled field is the low-pressure equilibrium field for basaltic magma (Fe/Mg Kd = 0·30± 0·03; Roeder & Emslie, 1970; Ulmer, 1989). Many of the olivines plot below the equilibrium field, especially for the lavas with higher mg-numbers, and so probably experienced olivine accumulation. The lavas from episode 54 (B and D vents) have the lowest mg-number but thelargest range in forsterite content. This large Fo range is probably the result of magma mixing.
olivines with forsterite contents too low to be in equi- contents ever reported for Hawaiian rocks (up to 97·6%;Table 3).librium with melts corresponding to their whole-rock
composition (Fig. 2). The olivines in these rocks areidentical in appearance to those in Pu’u ’O’o lavaswith lower mg-numbers. The positive correlation betweenolivine abundance and mg-number in Pu’u ’O’o lavas Clinopyroxene(Table 2) indicates that these lavas probably accumulated Compositions were determined for >70 cpx crystals inolivine. Sample 11-May-98 contains olivine crystals with 12 lavas from episodes 53 and 54 (Table 4). The crystalsforsterite contents up to >89%, which are too high to are compositionally unzoned or weakly normally zoned.be in equilibrium with the host rock mg-number. These Microphenocryst compositions in these lavas range inhigh-Fo crystals show no signs of deformation and are cpx end-member components from 10 to 17% ferrosilite,probably relics from a Pu’u ’O’o parental magma. 28 to 43% wollastonite and 47 to 53% enstatite. Matrix
Episode 54 olivines are more variable in composition, crystals have a wider range in iron content (e.g. 8–24%with cores ranging from 78 to 85% Fo. Most of the ferrosilite). Cr2O3 contents of Pu’u ’O’o cpx are from 0·1rare olivines in these lavas are too forsteritic to be in to 1·3 wt %, Al2O3 contents range from 1·3 to 3·8 wt %equilibrium with the differentiated composition of their and Na2O and TiO2 contents are low (<0·25 wt % andhost rock (Fig. 2), although they are strongly zoned with <1·5 wt %). These compositions are typical of cpx fromrims in equilibrium with their bulk composition. These Hawaiian tholeiites (Fodor et al., 1975). Many of thehigh-Fo olivines were probably derived from a more cpx crystals have low and variable CaO contents (e.g.MgO-rich magma shortly before eruption. 14·1–19·1 wt % for rim to core) but nearly constant
The black olivines in the basalt xenoliths from episode mg-numbers (80·4–81·1; Table 4). These features areindicative of disequilibrium growth (see Lofgren, 1980).51 and 53 lavas have some of the highest forsterite
973
JOURNAL OF PETROLOGY VOLUME 41 NUMBER 7 JULY 2000
Disequilibrium cpx was also noted in the early, evolved elements relative to a highly incompatible element (e.g.Pu’u ’O’o lavas but its presence was ascribed to magma K) define steep negative trends for episode 50–53 andmixing (Garcia et al., 1992). No petrographic or mineral 55 lavas that overlap with and extend beyond the fieldchemical evidence for magma mixing was observed in for older Pu’u ’O’o lavas (Fig. 4). There is a systematicthe episode 50–53 and 55 lavas. decrease in K at a given Ni abundance indicating a
progressive variation in parental magma compositionduring the eruption. Among the incompatible trace ele-ments, Rb and Nb varied by a factor of two for episode
Plagioclase 50–55 lavas, with less variation for Zr (1·65), Y (1·45),Compositions were determined for phenocrysts and mi- Sr (1·4) and V (1·3). Element–element and element-ratiocrophenocrysts in two episode 54 D vent samples (Table plots of the highly incompatible elements demonstrate5), which are the only Pu’u ’O’o lavas we have sampled that the episode 50–53 and 55 lavas overlap in com-since episode 3 to contain plagioclase phenocrysts. These position with previous Pu’u ’O’o lavas and thus wereplagioclase crystals are strongly reversely zoned (7–8% derived from compositionally similar mantle sources. Theanorthite) from core to near the rim (>30 �m) but are episode 54 lavas are similar to early Pu’u ’O’o lavas innormally zoned at the rim. Cr and Sr but have distinctly lower Ni and Nb for their
K concentration (Fig. 4). These differences indicate thatthe episode 54 lavas are probably not related to the samebatch of magma that was tapped during episodes 1–3
WHOLE-ROCK GEOCHEMISTRY despite the close spatial relationships of their vents. TheXRF major and trace element analyses relatively small Sr variation (Table 5) and the decrease
in Al2O3 with decreasing MgO for the most evolvedNinety-two samples from episode 50–55 lavas (>1·2rocks, especially episode 54 lavas (Fig. 3), indicates thatsamples/month over a>6·7 year period) were analyzedthe evolved lavas underwent plagioclase fractionation.by XRF spectrometry in duplicate at the University of
Massachusetts for major and trace elements (Rb, Sr, Y,Nb, Zr, Zn, Ni, Cr and V; Table 6). For details of themethods used and for analytical precision estimates, the
ICP-MS analysesreader is referred to Rhodes (1996). With these newTwenty-eight samples from episodes 50 to 55 were ana-analyses, our XRF dataset for Pu’u ’O’o lavas includeslyzed for trace elements by ICP-MS in the same laborat-395 samples (a full dataset of these analyses is availableory (Washington State University; Table 7) as the 32on the Journal of Petrology Web page, at http://www.lavas from episodes 1 to 48 (Garcia et al., 1996). For apetrology.oupjournals.org).summary of the methods used, the reader is referred toThe lavas from episodes 50 to 55 extend the com-King et al. (1993) and Pietruszka & Garcia (1999a). Thepositional range for the Pu’u ’O’o eruption (Fig. 3). Aanalytical precision (1�) is estimated to be 1–3% basedlava from episode 53 (19-Jan-96) is the most MgO richon repeated analyses of a Kilauea basalt standardof any Pu’u ’O’o lava (10·1 wt %) and an episode 54[Kil1919; for information on this standard, see Rhodeslava (Vent F) has the lowest MgO (5·6 wt %). Overall,(1996) for XRF data and Pietruszka & Garcia (1999a)the episode 50–55 lavas are similar in composition tofor ICP-MS data]. Plots of highly incompatible elementsthe other Pu’u ’O’o lavas, especially for Al2O3. In detail,show excellent linear trends for the overall Pu’u ’O’othe K2O and CaO trends are offset to lower values, withsuite with the episode 50–55 lavas extending the rangethe episode 55 lavas having the lowest values. Also, theof the suite to both significantly higher and lower con-K2O–MgO trend is linear for episode 50–53 and 55centrations (Fig. 5). Plots of ratios of highly over mod-lavas compared with the curved trend observed for theerately incompatible elements vs highly incompatibleearlier episodes of the eruption and expected from crystalelements (e.g. La/Yb vs Ba) show an overall linear trendfractionation (Fig. 3). This linear trend is probably relatedwith the scatter reflecting variable amounts of crystalto mantle melting processes, which are discussed below.fractionation. The episode 54 lavas have distinctly higherThe episode 54 lavas are petrographically and geo-La/Yb ratios compared with all previously analyzed Pu’uchemically similar to lavas from episodes 1 to 3 with low’O’o lavas except for an episode 1 lava (Fig. 5). OnMgO contents (5·6–6·3 wt %) and are readily dis-ratio–ratio plots of highly incompatible elements (withtinguished from other Pu’u ’O’o lavas by their relativelythe most incompatible element in the numerator; e.g.low CaO/Al2O3 (<0·78 vs >0·82).Ba/Ce and La/Ce) the episode 50–53 and 55 lavas formCr, Ni and Zn are compatible trace elements in thefields that overlap with the field for episode 1–48 lavasepisode 50–55 Pu’u ’O’o lavas (Table 5). Cr and Nibut extend to lower values for the more recent lavasshow the most variation of any element in Pu’u ’O’o
lavas (factors of 6·4 and 3·3, respectively). Plots of these (1995–1998; Fig. 5). The overall decrease in La/Yb with
974
GARCIA et al. MAGMATIC PROCESSES FOR A PROLONGED KILAUEA ERUPTION
Tab
le4
:R
epre
sent
ativ
em
icro
prob
ean
alys
esof
clin
opyr
oxen
esin
Pu’
u’O
’ola
vas
from
epis
odes
53
and
54
Ep
iso
de:
5353
5353
5354
Sam
ple
:9-
May
-93
18-M
ay-9
429
-Mar
-95
14-J
un
-95
14-O
ct-9
5D
ven
t
mp
h-c
mp
h-r
mp
h-c
mp
h-r
mp
h-c
mp
h-c
mp
h-c
mp
h-c
mp
h-r
mat
rix
mp
h-c
mat
rix
SiO
251
·34
53·5
951
·92
50·4
352
·47
51·3
852
·63
52·4
551
·75
50·3
552
·69
51·3
5
TiO
20·
920·
520·
911·
300·
540·
860·
580·
620·
801·
480·
541·
18
Al 2
O3
2·65
1·32
2·26
3·19
2·28
3·32
2·03
2·13
2·71
3·79
1·94
3·38
Cr 2
O3
0·24
0·11
0·20
0·17
0·80
1·09
0·74
0·74
1·01
0·53
0·74
0·32
FeO
7·49
8·64
8·05
8·88
6·24
6·25
6·21
6·44
6·54
7·70
6·29
8·18
Mn
O0·
160·
190·
170·
190·
330·
22—
0·18
0·17
0·19
——
Mg
O17
·26
20·8
117
·69
17·6
117
·98
16·6
817
·63
18·1
217
·10
15·8
317
·96
17·6
3
CaO
19·1
414
·09
18·0
817
·30
18·6
419
·62
19·7
119
·28
19·3
319
·29
18·8
816
·86
Na 2
O0·
400·
540·
370·
460·
220·
240·
210·
220·
200·
230·
200·
19
Tota
l99
·60
99·8
199
·65
99·5
399
·50
99·6
699
·74
100·
1899
·61
99·3
999
·24
99·1
0
Pyr
oxe
ne
end
mem
ber
s(m
ol%
)
En
49·0
58·1
50·2
50·3
51·5
48·6
50·3
50·9
50·9
46·5
51·2
51·3
Fs11
·913
·512
·814
·210
·010
·29·
310
·210
·212
·710
·113
·4
Wo
39·1
28·3
36·9
35·5
38·4
41·1
40·4
38·9
38·9
40·8
38·7
35·4
mg
-no
.80
·481
·179
·777
·983
·782
·683
·583
·482
·378
·683
·679
·3
Geo
ther
mo
bar
met
ry(P
utr
ika
etal
.,19
96)
T(°
C)
1212
1158
1206
1168
1190
1195
1196
1198
1209
1212
1210
1167
P(G
Pa)
0·60
−0·
080·
510·
230·
430·
550·
470·
420·
550·
680·
550·
20
mp
h,
mic
rop
hen
ocr
ysts
;c,
core
;r,
rim
;al
lo
xid
eco
nce
ntr
atio
ns
are
inw
t%
;p
yro
xen
em
g-n
o.
is[M
g/(
Fe+
Mg
)]×
100.
975
JOURNAL OF PETROLOGY VOLUME 41 NUMBER 7 JULY 2000
Table 5: Microprobe analyses of plagioclase in two samples from vent D, episode 54
Sample type and Area SiO2 Al2O3 FeO CaO Na2O Total An %
size
ph, 0·9 mm core 52·51 30·09 0·75 12·93 4·00 100·28 64·1
near rim 50·52 31·41 0·77 14·41 3·20 100·30 71·3
rim 51·07 30·50 0·76 13·82 3·46 99·61 68·8
mph, 0·3 mm core 50·42 31·57 0·65 14·49 3·30 100·43 70·8
near rim 48·35 32·65 0·71 15·88 2·39 99·97 78·6
rim 52·73 29·91 0·63 12·86 3·95 100·08 65·3
An %, mole per cent anorthite; ph, phenocryst; mph, microphenocryts.
time could be caused by an increase in the degree of 1999a), Pu’u ’O’o lavas have the lowest 206Pb/204Pb ratiosand plot within the Loihi field (Fig. 7). Nd isotope ratiospartial melting during the eruption. The decrease in
ratios of highly incompatible trace elements cannot be have not varied beyond analytical error for the entireeruption (Garcia et al., 1996; Table 7).explained by an increase in partial melting and instead
indicates that there was a minor change in the mantlesource composition for the more recent lavas.
MAGMATIC CONDITIONS FOR THEPU’U ’O’O ERUPTIONPb, Sr and Nd isotopesThe Pu’u ’O’o eruption provides an excellent opportunityRatios of Pb, Sr and Nd isotopes were determined onto examine the details of magmatic processes at an activeselected Pu’u ’O’o samples at the University of Hawaiibasaltic volcano because we have a wealth of backgroundusing a VG Sector mass spectrometer [see Pietruszka &information on Kilauea’s basic magmatic processes (e.g.Garcia (1999a) for a summary of methods used]. IsotopeTilling & Dvorak, 1993). Seismic evidence has docu-fractionation corrections, standard values, total pro-mented a vertical conduit of 60 km length that suppliescedural blanks, and analytical uncertainties are given inmagma to Kilauea (Tilling & Dvorak, 1993). For theTable 8. Five samples, one each from episodes 51 andPu’u ’O’o eruption, magma in this conduit is thought to55, and three from 53, were analyzed for isotope ratios;bypass the summit reservoir and feed directly into thethe data for two of these samples (29-Dec-92 and 25-east rift zone conduit (based on the rapid change in traceApr-94) were presented by Garcia et al. (1996). Ourelement geochemistry for Pu’u ’O’o lavas compared withprevious work showed that Pb and Sr isotope ratiosKilauea summit lavas; Garcia et al., 1996). Seismic andchanged somewhat during the first 2 years of the eruption,ground deformation studies indicate that the east rifta period of extensive magma mixing and crustal con-zone supplies a shallow (>0·5–2·5 km depth) magmatamination (Garcia et al., 1992, 1998a), but remainedreservoir under the Pu’u ’O’o vent (Koyanagi et al., 1988;nearly constant for the next 7 years (Garcia et al., 1996).Hoffmann et al., 1990; Gillard et al., 1996). From 1983The new isotopic data for episode 51, 53 and 55 lavasto 1986, this reservoir was a narrow (>3 m wide), dike-show slight temporal increases in Pb and Sr isotope
ratios, although the Sr isotope variation is essentially like body (>1·6 km long and 2·5 km deep) that belchedout much of its contents during brief (>1 day), monthlywithin analytical error (Fig. 6). Pu’u ’O’o lavas deviate
from the good negative correlation observed for Pb and eruptions (Hoffmann et al., 1990). The crown of the Pu’u’O’o magma reservoir is open, which allows heat andSr isotope ratios in most Hawaiian tholeiites (e.g. West
et al., 1987), as do Loihi tholeiites (Fig. 7). The Pb and gases to escape readily (Wolfe et al., 1988). The highsurface area and open top of this magma reservoir isSr isotopic data for tholeiites erupted during the last
>20 years from the three adjacent, active Hawaiian thought to have caused rapid crystal fractionation (>5%)during the 20–30 day pauses between episodes 4–47volcanoes are distinctly different (Fig. 7) and demonstrate
that the Hawaiian plume source must have at least three (Garcia et al., 1992). There are no data to constrain thecurrent size or shape of the Pu’u ’O’o reservoir sincedistinct components [as suggested by Staudigel et al.
(1984)]. Compared with other historical Kilauea lavas episode 47 but it is assumed to have remained small (e.g.Heliker et al., 1998b). Its shape may have changed to aanalyzed in the same laboratory (Pietruszka & Garcia,
976
GARCIA et al. MAGMATIC PROCESSES FOR A PROLONGED KILAUEA ERUPTION
Tab
le6
:X
RF
who
le-r
ock
anal
yses
ofP
u’u
’O’o
lava
sfr
omep
isod
es5
0to
55
Sam
ple
Day
SiO
2Ti
O2
Al 2
O3
Fe2O
3M
nO
Mg
OC
aON
a 2O
K2O
P2O
5To
tal
YS
rR
bN
bZ
rN
iC
rV
Zn
Ep
iso
de
50
18-F
eb-9
233
3349
·78
2·37
012
·79
12·4
00·
178·
7510
·59
2·14
0·41
40·
223
99·6
324
·131
76·
812
·915
116
752
129
111
2
19-F
eb-9
233
3449
·75
2·36
812
·82
12·3
00·
188·
7810
·55
2·32
0·42
70·
226
99·7
224
·332
07·
013
·215
117
050
728
411
3
22-F
eb-9
233
3749
·75
2·33
512
·67
12·4
50·
179·
3010
·45
2·18
0·41
10·
221
99·9
424
·431
56·
513
·115
117
153
928
912
9
23-F
eb-9
2a33
3849
·76
2·33
812
·71
12·4
40·
179·
1610
·47
2·21
0·40
90·
221
99·8
924
·331
56·
713
·014
718
555
728
911
8
23-F
eb-9
2b33
3849
·52
2·32
512
·63
12·3
70·
179·
0910
·42
2·17
0·49
60·
230
99·4
224
·031
36·
912
·714
718
456
728
113
4
Ep
iso
de
51
12-M
ar-9
233
5649
·68
2·35
012
·71
12·4
40·
188·
9710
·53
2·12
0·41
20·
221
99·6
124
·431
66·
813
·014
817
654
128
611
9
29-M
ar-9
233
7349
·88
2·36
612
·84
12·4
30·
178·
7610
·59
2·23
0·42
30·
224
99·9
1—
——
13·1
148
167
517
282
116
05-A
pr-
9233
8049
·76
2·36
812
·71
12·4
10·
188·
8710
·55
2·25
0·42
20·
230
99·7
524
·632
16·
613
·814
916
952
528
613
0
12-A
pr-
9233
8749
·65
2·35
112
·64
12·3
60·
188·
9910
·48
2·13
0·42
10·
230
99·4
323
·931
86·
913
·814
717
854
428
416
1
27-A
pr-
9234
0249
·74
2·36
312
·75
12·3
80·
188·
7710
·54
2·22
0·42
10·
230
99·5
924
·332
37·
114
·014
916
452
527
311
7
26-M
ay-9
234
3149
·58
2·33
912
·65
12·5
40·
189·
0410
·51
2·07
0·41
30·
220
99·5
424
·131
57·
013
·814
717
654
627
511
7
06-J
un
-92
3442
49·6
52·
352
12·6
812
·49
0·18
8·94
10·5
92·
110·
416
0·23
099
·64
23·9
319
7·1
14·7
149
172
530
282
118
13-J
un
-92
3449
49·6
32·
345
12·6
812
·37
0·18
8·86
10·5
32·
200·
419
0·23
099
·44
24·3
318
7·1
13·9
151
173
522
283
134
16-J
un
-92
3452
49·6
92·
363
12·6
912
·42
0·18
8·80
10·5
32·
240·
420
0·23
099
·56
24·1
322
6·9
13·7
153
172
516
280
121
07-J
ul-
9234
7349
·76
2·38
612
·85
12·3
40·
188·
5110
·59
2·27
0·42
40·
230
99·5
424
·631
97·
113
·715
216
248
428
712
1
27-A
ug
-92
3524
49·8
82·
397
13·0
312
·33
0·18
8·33
10·6
82·
310·
427
0·23
099
·79
24·5
324
7·3
13·6
154
142
463
288
121
08-S
ep-9
235
3649
·88
2·36
412
·84
12·3
60·
188·
6210
·54
2·27
0·42
00·
230
99·7
024
·432
06·
613
·415
115
748
428
512
1
05-O
ct-9
235
6350
·27
2·44
513
·18
12·3
20·
187·
9310
·78
2·37
0·43
40·
240
100·
1525
·032
87·
114
·015
313
139
429
011
9
01-N
ov-
9235
9049
·69
2·34
112
·60
12·4
60·
189·
0910
·51
2·24
0·42
60·
220
99·7
624
·131
56·
712
·614
417
451
527
911
7
07-N
ov-
9235
9649
·63
2·35
012
·63
12·5
10·
188·
9510
·49
2·17
0·41
80·
230
99·5
624
·832
06·
713
·114
716
148
828
311
9
12-N
ov-
9236
0149
·81
2·38
712
·82
12·3
80·
188·
5210
·61
2·20
0·42
30·
230
99·5
624
·632
07·
513
·114
616
048
228
413
5
29-D
ec-9
236
4849
·57
2·35
512
·69
12·4
20·
188·
8010
·52
2·22
0·41
90·
230
99·4
024
·031
86·
813
·514
917
049
828
012
0
09-J
an-9
336
5949
·81
2·37
312
·84
12·4
00·
188·
5710
·63
2·29
0·42
30·
230
99·7
524
·332
07·
013
·314
914
747
428
336
4
Ep
iso
de
52
04-O
ct-9
235
6249
·74
2·35
912
·79
12·3
70·
188·
7310
·55
2·31
0·41
90·
230
99·6
824
·631
96·
713
·215
015
648
428
012
0
Ep
iso
de
53
20-F
eb-9
337
0150
·00
2·40
412
·97
12·4
50·
188·
3710
·71
2·26
0·41
50·
232
99·9
924
·431
77·
213
150
132
452
282
120
18-M
ar-9
337
2749
·93
2·38
212
·87
12·5
50·
188·
6410
·63
2·20
0·41
50·
231
100·
0324
·131
37·
112
·814
814
049
328
111
6
16-A
pr-
9337
5649
·86
2·37
912
·85
12·4
90·
178·
5610
·64
2·14
0·41
30·
231
99·7
324
·131
37·
112
·814
814
350
128
411
6
09-M
ay-9
337
7949
·77
2·34
712
·73
12·4
70·
188·
9010
·55
2·23
0·41
90·
230
99·8
324
·231
76·
813
·314
916
647
628
512
1
10-M
ay-9
337
8049
·91
2·30
312
·80
12·4
90·
188·
6210
·64
2·20
0·40
90·
227
99·8
325
·131
96·
513
·914
915
150
428
211
7
977
JOURNAL OF PETROLOGY VOLUME 41 NUMBER 7 JULY 2000
Tab
le6
:co
ntin
ued
Dat
eD
ayS
iO2
TiO
2A
l 2O
3Fe
2O3
Mn
OM
gO
CaO
Na 2
OK
2OP
2O5
Tota
lY
Sr
Rb
Nb
Zr
Ni
Cr
VZ
n
09-J
un
-93
3810
49·7
82·
370
12·7
612
·48
0·18
8·58
10·6
22·
280·
488
0·23
199
·77
24·3
314
7·3
12·9
148
139
490
284
116
08-J
ul-
9338
3949
·96
2·38
112
·79
12·5
20·
188·
5610
·64
2·35
0·41
30·
231
100·
0124
·331
37·
113
·814
814
050
328
311
34
13-A
ug
-93
3875
50·1
22·
390
12·9
612
·51
0·18
8·41
10·7
62·
140·
418
0·23
110
0·12
24·8
321
6·5
14·3
146
140
458
290
160
27-S
ep-9
339
2050
·15
2·37
912
·96
12·4
70·
188·
4010
·75
2·09
0·41
70·
232
100·
0324
·832
17·
214
·614
413
545
328
911
6
30-O
ct-9
339
5350
·35
2·46
613
·35
12·3
10·
187·
2611
·07
2·24
0·43
20·
239
99·9
0—
331
7·1
15·1
152
9437
629
518
2
14-D
ec-9
339
9849
·87
2·38
812
·99
12·4
10·
188·
1910
·79
2·36
0·42
00·
230
99·8
324
·731
96·
713
·815
413
044
129
512
0
04-J
an-9
440
1950
·38
2·47
013
·41
12·3
30·
187·
2811
·04
2·38
0·43
60·
240
100·
1525
·433
16·
914
·715
797
367
298
169
11-F
eb-9
440
5750
·02
2·40
213
·04
12·4
50·
188·
1510
·82
2·31
0·42
40·
230
100·
0325
·032
37·
114
·315
711
943
929
711
9
13-M
ar-9
440
8749
·99
2·40
113
·02
12·4
50·
188·
1710
·81
2·32
0·42
30·
230
99·9
925
·132
37·
513
·915
612
844
129
711
9
09-A
pr-
9441
1450
·38
2·44
013
·26
12·4
00·
187·
8010
·92
2·41
0·42
70·
240
100·
4625
·332
67·
613
·815
811
040
930
411
8
11-A
pr-
9441
1650
·05
2·38
013
·02
12·4
60·
188·
3410
·78
2·34
0·42
20·
230
100·
2024
·432
16·
514
·415
514
244
029
411
9
25-A
pr-
9441
3050
·16
2·44
113
·27
12·4
20·
187·
7110
·93
2·45
0·42
90·
240
100·
2325
·732
96·
614
·115
610
737
930
012
0
18-M
ay-9
441
5350
·01
2·45
213
·23
12·3
60·
187·
6110
·91
2·33
0·43
00·
238
99·8
124
·832
67·
513
·415
310
438
329
011
6
21-J
un
-94
4187
50·2
02·
437
13·3
312
·42
0·18
7·89
10·9
32·
390·
430
0·24
010
0·45
24·9
326
7·6
13·9
157
127
414
296
146
11-J
ul-
9442
0750
·05
2·42
613
·05
12·4
10·
187·
9310
·84
2·26
0·42
50·
233
99·8
024
·832
37·
513
·315
211
550
929
011
7
02-A
ug
-94
4229
49·9
92·
434
13·1
312
·40
0·18
7·72
10·8
52·
360·
433
0·24
099
·74
24·9
314
7·4
13·2
153
109
396
285
117
25-A
ug
-94
4252
49·8
32·
400
12·9
612
·44
0·18
8·00
10·7
92·
310·
427
0·23
099
·57
24·4
305
7·0
13·2
151
119
409
284
116
09-O
ct-9
442
9749
·82
2·38
612
·91
12·5
10·
188·
1910
·72
2·23
0·40
90·
230
99·5
925
·532
57·
114
·115
012
442
829
212
0
16-N
ov-
9443
3550
·14
2·40
913
·31
12·5
10·
188·
0510
·78
2·23
0·37
90·
250
99·8
625
·232
87·
214
·115
112
241
829
412
1
29-D
ec-9
443
7850
·02
2·41
513
·00
12·5
10·
188·
0310
·74
2·28
0·41
80·
240
99·8
324
·832
66·
914
·715
211
941
029
212
0
14-J
an-9
543
9449
·91
2·39
812
·96
12·5
60·
188·
2110
·71
2·22
0·40
60·
230
99·7
823
·732
56·
514
·015
012
944
329
212
0
24-F
eb-9
544
3550
·15
2·43
213
·14
12·5
00·
187·
8710
·80
2·17
0·42
30·
233
99·9
025
·232
57·
414
·215
212
041
229
511
9
29-M
ar-9
544
6850
·13
2·41
213
·10
12·4
90·
187·
9910
·76
2·17
0·42
10·
232
99·8
925
·132
36·
814
·315
211
742
329
312
0
27-A
pr-
9544
9749
·92
2·37
412
·91
12·5
70·
188·
4010
·63
2·15
0·41
10·
230
99·7
824
·531
96·
814
·414
813
145
528
711
8
13-J
un
-95
4544
49·7
42·
323
12·5
712
·79
0·18
9·21
10·4
22·
120·
404
0·22
499
·99
23·7
306
7·0
12·5
145
144
480
272
117
05-J
ul-
9545
6649
·87
2·40
312
·98
12·4
60·
187·
9710
·73
2·65
0·50
40·
233
99·4
824
·531
77·
212
·715
010
941
528
211
6
15-A
ug
-95
4607
49·8
72·
375
12·9
012
·56
0·18
8·32
10·6
42·
060·
412
0·22
699
·54
24·4
314
7·2
12·7
149
125
453
284
117
11-S
ep-9
546
3450
·39
2·46
013
·40
12·3
00·
187·
1511
·04
2·17
0·42
70·
234
99·7
524
·832
37·
312
·915
284
382
288
115
14-O
ct-9
546
6749
·61
2·33
612
·63
12·8
20·
198·
9410
·38
1·95
0·40
60·
223
99·4
923
·830
76·
912
·514
614
644
617
711
8
19-J
an-9
647
6449
·32
2·23
412
·17
12·9
00·
1810
·14
10·0
51·
870·
386
0·21
499
·48
23·0
294
6·6
11·9
139
183
568
266
117
19-F
eb-9
647
9550
·02
2·37
312
·95
12·4
80·
188·
1410
·71
2·04
0·40
70·
226
99·5
424
·431
16·
812
·614
711
845
128
411
7
15-M
ar-9
648
2049
·99
2·35
612
·84
12·5
30·
188·
2710
·64
2·07
0·41
70·
224
99·5
224
·231
06·
912
·514
712
145
328
211
7
07-J
un
-96
4904
50·5
02·
405
13·3
512
·35
0·19
7·75
10·9
12·
210·
413
0·22
910
0·30
24·4
316
7·8
12·5
149
8136
628
311
4
22-A
ug
-96
4980
50·1
52·
369
13·0
312
·49
0·19
8·23
10·7
62·
240·
407
0·22
510
0·08
24·1
309
7·2
12·2
146
108
411
280
115
10-J
an-9
751
2150
·10
2·34
912
·98
12·4
60·
198·
3310
·74
2·08
0·40
20·
220
99·8
623
·930
77·
012
·214
312
546
828
412
5
26-J
an-9
751
3750
·18
2·30
512
·74
12·4
80·
189·
0310
·49
2·21
0·40
30·
218
100·
2423
·429
97·
012
140
145
523
274
114
30-J
an-9
751
4149
·85
2·34
312
·91
12·4
80·
198·
3210
·71
2·16
0·40
40·
220
99·5
924
·631
37·
212
·914
693
387
288
119
30-J
an-9
751
4150
·45
2·42
213
·31
12·3
30·
197·
5310
·98
2·24
0·41
80·
230
100·
1024
·731
47·
012
·514
696
377
280
114
978
GARCIA et al. MAGMATIC PROCESSES FOR A PROLONGED KILAUEA ERUPTION
Dat
eD
ayS
iO2
TiO
2A
l 2O
3Fe
2O3
Mn
OM
gO
CaO
Na 2
OK
2OP
2O5
Tota
lY
Sr
Rb
Nb
Zr
Ni
Cr
VZ
n
Ep
iso
de
54
Ven
tA
5141
50·2
12·
960
13·6
212
·63
0·19
6·37
10·3
52·
380·
588
0·31
099
·61
28·5
376
10·6
18·7
194
8215
432
012
5
Ven
tB
5141
50·1
82·
995
13·6
612
·62
0·19
6·15
10·3
42·
470·
604
0·32
099
·52
28·7
382
10·7
19·0
197
7711
431
912
5
Ven
tB
251
4150
·17
2·96
513
·63
12·6
10·
196·
3110
·33
2·46
0·59
00·
310
99·5
728
·237
910
·718
·719
481
138
313
123
Ven
tD
5141
50·4
12·
933
13·7
912
·54
0·19
6·30
10·4
32·
470·
635
0·31
010
0·01
28·2
375
10·4
18·3
191
7912
831
712
4
Ven
tE
5141
50·1
52·
970
13·7
212
·61
0·19
6·22
10·3
52·
490·
594
0·31
099
·61
28·5
378
10·7
18·7
195
7913
032
212
4
Ven
tF
5141
50·9
03·
443
13·5
013
·16
0·18
5·61
9·63
2·62
0·72
30·
391
100·
1633
·638
813
·223
·323
868
8834
913
3
Ep
iso
de
55
04-A
pr-
9752
0550
·24
2·41
013
·28
12·2
50·
177·
5210
·90
2·36
0·41
00·
228
99·7
724
·331
47·
412
·514
895
366
285
116
21-A
pr-
9752
2250
·34
2·40
413
·29
12·3
80·
197·
5510
·93
2·19
0·41
60·
230
99·9
224
·731
47·
012
·514
696
377
280
114
25-M
ay-9
752
5650
·33
2·39
213
·21
12·3
90·
197·
7110
·91
2·14
0·41
30·
230
99·9
124
·531
27·
212
·614
610
338
728
511
5
13-J
ul-
9753
0550
·14
2·35
413
·08
12·3
80·
197·
9210
·85
2·15
0·40
40·
220
99·6
924
·330
96·
712
·414
310
137
727
611
5
23-J
ul-
9753
1550
·39
2·38
013
·20
12·3
50·
197·
6010
·96
2·13
0·40
80·
230
99·8
424
·431
17·
012
·414
595
381
282
114
10-A
ug
-97
5333
49·5
62·
245
12·4
212
·72
0·18
9·84
10·1
72·
260·
385
0·21
299
·99
22·9
293
6·8
11·7
138
177
528
267
116
11-O
ct-9
753
9550
·47
2·38
213
·16
12·2
60·
187·
9110
·86
2·24
0·40
70·
228
100·
1024
·131
06·
912
·414
598
413
272
112
18-O
ct-9
754
0250
·19
2·30
712
·85
12·5
70·
188·
8910
·53
2·21
0·39
40·
220
100·
3423
·229
96·
711
·814
014
752
126
211
2
26-D
ec-9
754
7150
·18
2·30
512
·74
12·4
80·
189·
0310
·49
2·21
0·40
30·
218
100·
2423
·429
97·
012
140
145
523
274
114
10-J
an-9
854
8650
·16
2·29
912
·83
12·4
80·
188·
9910
·47
2·10
0·39
30·
218
100·
1223
·429
96·
611
·814
015
054
027
432
1
17-J
an-9
854
9350
·16
2·28
212
·70
12·4
80·
189·
2610
·42
2·21
0·40
20·
215
100·
3123
·229
77·
011
·713
914
651
927
311
5
18-J
an-9
8E
5494
50·2
02·
291
12·7
712
·49
0·18
8·98
10·4
12·
150·
394
0·21
910
0·06
23·1
297
6·9
11·7
140
155
548
273
114
18-J
an-9
8S
5494
50·1
82·
301
12·8
212
·38
0·17
8·91
10·4
22·
160·
397
0·21
699
·95
23·3
298
7·0
11·8
140
153
552
274
114
21-J
an-9
854
9750
·66
2·37
113
·24
12·2
40·
177·
9610
·82
2·21
0·41
20·
224
100·
3124
·130
77·
112
·114
310
945
427
511
3
27-M
ar-9
855
6250
·07
2·31
912
·77
12·4
00·
179·
0310
·45
2·15
0·40
90·
221
99·9
923
·630
27·
212
·114
215
754
127
511
5
10-A
pr-
9855
7650
·09
2·32
612
·78
12·4
10·
178·
9410
·47
2·11
0·40
00·
221
99·9
223
·530
37·
112
·214
314
752
027
511
3
11-M
ay-9
856
0750
·21
2·31
612
·80
12·4
00·
179·
1010
·40
2·10
0·40
50·
221
100·
1323
·530
17·
112
·214
216
854
927
411
5
22-M
ay-9
856
1850
·29
2·35
812
·97
12·3
10·
178·
5610
·59
2·19
0·41
10·
225
100·
0724
·030
67·
212
·414
413
849
027
811
5
17-A
ug
-98
5705
50·3
42·
366
12·9
412
·35
0·18
8·70
10·5
82·
150·
410
0·22
610
0·24
23·6
307
7·4
12·4
145
141
476
227
115
07-S
ep-9
857
2650
·26
2·35
312
·98
12·3
20·
178·
6510
·58
2·24
0·40
60·
224
100·
1623
·730
67·
512
·414
514
549
327
811
5
Day
isth
en
um
ber
of
day
ssi
nce
the
star
to
fth
eP
u’u
’O’o
eru
pti
on
on
Jan
.03
,19
83.
Oxi
de
con
cen
trat
ion
sar
ein
wt
%;
trac
eel
emen
tsar
ein
pp
m.
979
JOURNAL OF PETROLOGY VOLUME 41 NUMBER 7 JULY 2000
Fig. 3. Whole-rock MgO variation diagrams for lavas from episodes 1 to 55 of the Pu’u ’O’o eruption. The overall curved (K2O, TiO2) andkinked (Al2O3, CaO) trends indicate that crystal fractionation was a dominant process controlling compositional variation for most Pu’u ’O’olavas. Some of the lavas from episodes 3 to 10 plot below the kink on the CaO diagram marking the onset of clinopyroxene fractionation. Theselavas show petrographic signs of magma mixing (Garcia et al., 1992). The scatter on these plots is greater than analytical error (the 2� error isabout symbol size), which indicates that the parental magma composition changed during the eruption. All values are in wt %. The two fieldsenclose the mafic lavas from episodes 50 to 53 and 55 and the evolved lavas from episode 54. The data are from Table 5 and Garcia et al.(1992, 1996).
more thermally efficient form with the change from Crystallization pressures for the Pu’u ’O’o reservoircan be inferred from several lines of evidence. Seismic andepisodic to continuous effusion in mid-1986.
The temperature of Pu’u ’O’o magma just before ground deformation studies indicates that the reservoir isshallow (<3 km depth) corresponding to pressures oferuption can be inferred from the MgO contents of glasses
(Helz & Thornber, 1987), clinopyroxene compositions <0·10 GPa (Hoffmann et al., 1990). Thermobarometrycalculations for cpx microphenocryst cores yield much(Putirka et al., 1996) and from thermocouple meas-
urements of molten lava. Temperatures of 1145–1160°C higher pressures for episode 53 magmas (>0·51± 0·09GPa; Table 4). Similar values were also reported forhave been determined for Pu’u ’O’o glasses (Mangan et
al., 1995; Heliker et al., 1998b), which agree well with cpx microphenocrysts in lavas from episodes 9 and 10(Putirka, 1997). Surprisingly, high pressures were alsomeasured lava temperatures (1135–1160°C). Tem-
peratures calculated from cpx matrix and micro- calculated for some matrix and microphenocryst rimcompositions (0·55 and 0·68 GPa; Table 4). These mantlephenocryst rim compositions yield similar to higher
temperatures (1158–1212°C; Table 4). Cpx micro- depth estimates (the Moho is >13 km, >4 GPa, at itsdeepest point under Kilauea; Klein et al., 1987) arephenocryst cores yield consistently high temperatures
(1187–1212°C), which might be considered more rep- inconsistent with petrographic, field, experimental andgeochemical evidence, which indicates that cpx formsresentative of magmatic rather than eruption tem-
peratures. Experiments on Kilauea lava compositions late in Pu’u ’O’o lavas.The liquidus mineralogy of a MgO-rich Pu’u ’O’o lavaindicate that lower temperatures (1160–1172± 9°C) are
more appropriate for low-pressure growth of cpx (Helz was modeled using the MELTS program (Ghiorso &Sack, 1995) to better understand the cause of the& Thornber, 1987).
980
GARCIA et al. MAGMATIC PROCESSES FOR A PROLONGED KILAUEA ERUPTION
black olivines; Table 3) is consistent with results of theMELTS calculations that olivine formed at low pressures(e.g. Ulmer, 1989). Thus, it is our conclusion that theminerals in the Pu’u ’O’o lavas record only the effectsof low-pressure processes (<0·3 GPa).
If the shallow magmatic processes in the rift zoneand Pu’u ’O’o reservoir can be identified and theircompositional effects removed from the geochemistry ofPu’u ’O’o lavas, then the geochemical signature of mantlemelting and source heterogeneity can be identified. Inthe following sections, we present an analysis of thecrustal and mantle magmatic processes affecting Pu’u’O’o lavas.
CRUSTAL MAGMATIC PROCESSESFOR THE PU’U ’O’O ERUPTIONOlivine fractionation and accumulationThe importance of olivine in controlling compositionalvariation in Hawaiian tholeiitic lavas has been recognizedfor many years (e.g. Powers, 1955). This observation isreaffirmed in Pu’u ’O’o lavas by the presence of onlynormally zoned, olivine phenocrysts and the relativelyhigh and nearly constant whole-rock CaO/Al2O3 (>0·83)in all Pu’u ’O’o lavas erupted since episode 20 in 1984(except those from episode 54). In addition to olivine
Fig. 4. Variation of Nb and Ni vs K (all in ppm) for Pu’u ’O’o lavas fractionation, olivine accumulation is important in these(fields and symbols are as in Fig. 3 except the episode 50–53 and 55 weakly to moderately olivine-phyric lavas. For example,lavas have been subdivided by date of eruption, 1992–1994 and during a 4 month period of episode 53, the lavas exhibit1995–1998; data from Table 5). The shift towards lower incompatible
the largest MgO variation (7·1–10·1 wt %) since theelement abundance with time (except for episode 54 lavas) and thelarge variation in Ni at a given K concentration relative to analytical onset of continuous effusion in 1986 but have essentiallyerror (about symbol size) should be noted. All values are in ppm. the same olivine composition (81·0 ± 0·5% Fo). The
overall MgO variation for the two extreme lavas fromthis period (samples 11-Sep-95, which has olivine in
anomalous cpx pressure estimates. It was assumed for equilibrium with a melt of its whole-rock composition,this modeling that the magma underwent equilibrium and 19-Jan-96, the most MgO-rich lava from the entirecrystallization, contained 0·3 wt % H2O and 0·1 wt % eruption) can be explained by accumulation of>6·9 vol.CO2, and that its redox state was one log unit below the % olivine with 81% Fo. This result is consistent with theFMQ (fayalite–magnetite–quartz) buffer (models were relatively large difference in olivine abundance in thesealso run at FMQ with no significant change in the results). two samples (5·8± 1·5 vol. %). Good agreement betweenThe model results indicate that olivine is the liquidus predicted and observed olivine differences is also foundphase only at low pressures (<0·3 GPa). At moderate for the less MgO-rich sample 14-Oct-95 (8·9 wt %) andpressures (0·3–0·6 GPa), orthopyroxene is the liquidus sample 11-Sep-95 (predicted 3·9 vol. % vs observed 3·6phase, rather than cpx, and olivine is unstable. Euhedral ± 0·8 vol. %). These results are consistent with theolivine is ubiquitous in all but the most evolved Pu’u observation that most of the MgO-rich, episode 50–53’O’o lavas and orthopyroxene has not been observed lavas have olivine forsterite contents too low to be inin any of the Pu’u ’O’o lavas. Previous studies have equilibrium with their host rocks (Fig. 2). Thus, theconsistently shown that cpx phenocrysts are rare in episode 50–53 melts probably accumulated olivine andKilauea lavas, except those with MgO contents <6·8 wt were less MgO rich than the episode 48 magmas.% (e.g. lavas with>5 wt % MgO from the 1955 eruptioncontain cpx; Macdonald & Eaton, 1964). Therefore, the
Crustal assimilationcalculated moderate pressure for cpx crystallization inPu’u ’O’o magmas is probably an artifact of their rapid Pu’u ’O’o lavas have relatively low matrix oxygen isotopegrowth (as discussed above). The moderate CaO content ratios (4·6–5·2‰) and the oxygen isotope ratios for oli-
vines in many of these lavas are out of equilibrium withof the Pu’u ’O’o olivines (0·25–0·32 wt %, except in the
981
JOURNAL OF PETROLOGY VOLUME 41 NUMBER 7 JULY 2000
Tab
le7
:IC
P-M
San
alys
esof
sele
cted
Pu’
u’O
’ola
vas
from
epis
odes
50
to5
5
Ep
iso
de
Sam
ple
Day
Rb
Cs
Ba
LaC
eP
rN
dS
mE
uG
dT
bD
yH
oE
rT
mY
bLu
Th
YN
bH
fU
5112
-Mar
-92
3356
7·5
0·07
210
011
·428
·54·
2919
·65·
251·
805·
610·
865·
180·
972·
550·
352·
040·
290·
8727
·812
·23·
710·
295
5106
-Ju
n-9
234
427·
70·
078
106
12·1
29·9
4·42
20·7
5·61
1·95
5·99
0·92
5·42
1·02
2·68
0·38
2·18
0·32
0·88
28·8
12·4
3·99
0·30
1
5108
-Sep
-92
3536
7·1
0·07
199
11·4
28·5
4·21
19·5
5·16
1·83
5·60
0·84
5·04
0·94
2·46
0·35
1·98
0·28
0·82
28·0
12·6
3·69
0·28
1
5112
-No
v-92
3601
7·8
0·07
510
411
·930
·34·
4820
·45·
571·
925·
830·
915·
401·
042·
730·
392·
160·
310·
8528
·412
·43·
890·
297
5129
-Dec
-92
3648
7·2
0·07
210
311
·729
·04·
3520
·05·
341·
885·
960·
885·
190·
982·
580·
372·
050·
290·
8728
·412
·83·
840·
293
5320
-Feb
-93
3701
7·6
0·06
999
11·3
28·7
4·33
19·9
5·28
1·83
5·76
0·86
5·05
0·97
2·53
0·35
2·06
0·29
0·90
27·4
12·4
3·79
0·28
6
5309
-May
-93
3779
7·3
0·06
699
11·2
28·1
4·15
19·1
5·09
1·81
5·57
0·83
4·92
0·93
2·47
0·34
2·00
0·28
0·81
26·8
12·6
3·73
0·28
4
5313
-Au
g-9
338
757·
90·
072
108
11·9
30·3
4·41
20·3
5·64
1·93
5·93
0·89
5·38
1·03
2·66
0·37
2·16
0·30
0·92
28·3
13·2
4·05
0·31
6
5304
-Jan
-94
4019
7·9
0·07
510
611
·829
·94·
4220
·75·
451·
905·
940·
885·
340·
992·
570·
372·
120·
300·
9028
·512
·93·
880·
321
5325
-Ap
r-94
4130
8·1
0·07
610
912
·330
·84·
6321
·45·
731·
976·
210·
915·
581·
042·
770·
372·
210·
300·
9229
·313
·34·
050·
320
5325
-Au
g-9
442
527·
80·
072
102
11·5
28·6
4·29
19·6
5·29
1·84
5·88
0·86
5·22
0·98
2·57
0·36
2·11
0·29
0·84
27·9
13·3
3·90
0·29
2
5309
-Oct
-94
4297
7·5
0·07
110
411
·729
·34·
3220
·15·
351·
895·
790·
875·
090·
982·
550·
362·
080·
300·
8527
·013
·13·
880·
300
5314
-Jan
-95
4394
7·5
0·07
210
211
·429
·34·
3020
·15·
461·
855·
760·
875·
210·
952·
610·
362·
100·
290·
9027
·513
·03·
880·
304
5327
-Ap
r-95
4497
7·6
0·07
598
11·1
28·0
4·20
19·0
5·19
1·87
5·58
0·88
5·03
0·95
2·49
0·35
2·12
0·29
0·89
27·8
13·0
3·75
0·28
8
5315
-Au
g-9
546
077·
20·
071
9811
·328
·94·
2719
·55·
211·
805·
750·
865·
130·
992·
590·
362·
100·
290·
8527
·512
·93·
900·
290
5314
-Oct
-95
4667
7·4
0·06
899
11·0
27·5
4·06
18·6
5·18
1·79
5·67
0·85
5·03
0·95
2·50
0·35
2·02
0·28
0·85
27·3
12·7
3·74
0·29
1
5319
-Jan
-96
4764
6·7
0·06
291
10·4
26·3
3·91
18·1
4·92
1·69
5·19
0·80
4·76
0·90
2·43
0·33
1·91
0·27
0·81
25·7
11·8
3·56
0·27
9
5315
-Mar
-96
4820
7·2
0·06
797
11·0
27·7
4·17
18·9
5·15
1·78
5·71
0·85
5·19
0·98
2·56
0·36
2·12
0·29
0·83
27·2
12·6
3·88
0·29
0
5307
-Ju
n-9
649
047·
20·
071
100
11·4
28·8
4·27
19·9
5·33
1·90
5·66
0·87
5·26
0·99
2·61
0·35
2·12
0·29
0·84
27·3
12·4
3·88
0·27
2
5322
-Au
g-9
649
807·
60·
072
101
11·6
29·0
4·33
19·8
5·32
1·86
5·72
0·86
5·19
0·99
2·55
0·36
2·08
0·29
0·93
29·0
12·6
3·80
0·31
7
5310
-Jan
-97
5121
7·4
0·06
594
10·7
26·7
4·01
18·3
4·95
1·74
5·44
0·81
4·89
0·93
2·44
0·34
1·98
0·28
0·84
27·8
12·8
3·58
0·28
4
54Ve
nt
F51
4113
·40·
129
181
20·5
49·9
7·31
31·6
7·95
2·66
8·32
1·24
7·25
1·36
3·54
0·49
2·91
0·39
1·57
37·1
23·2
6·09
0·55
3
54Ve
nt
B51
4111
·00·
100
151
16·9
41·4
6·04
27·3
6·76
2·29
7·10
1·04
6·06
1·15
2·99
0·41
2·42
0·34
1·27
32·3
19·0
5·09
0·43
0
54Ve
nt
E51
4111
·30·
102
148
16·2
40·0
5·81
26·0
6·62
2·25
7·02
1·03
5·98
1·12
2·94
0·41
2·40
0·33
1·28
32·4
18·6
4·89
0·43
3
5530
-Jan
-97
5141
7·5
0·07
510
011
·629
·34·
3819
·95·
461·
915·
810·
885·
241·
012·
620·
372·
170·
300·
8928
·212
·53·
840·
302
5521
-Ap
r-97
5222
7·4
0·06
910
111
·428
·84·
1519
·65·
311·
895·
900·
875·
230·
982·
560·
362·
140·
300·
8427
·913
·03·
880·
285
5523
-Ju
l-97
5315
7·2
0·06
897
10·9
27·3
4·13
19·3
5·17
1·79
5·65
0·85
4·99
0·95
2·53
0·35
2·06
0·29
0·81
28·1
12·7
3·75
0·27
4
5510
-Jan
-98
5486
6·9
0·06
592
10·4
26·5
3·96
18·2
4·91
1·74
5·42
0·81
4·81
0·91
2·38
0·34
1·97
0·27
0·75
26·5
11·6
3·58
0·26
6
Day
isth
en
um
ber
of
day
ssi
nce
the
star
to
fth
eP
u’u
’O’o
eru
pti
on
(Jan
.03
,19
83);
all
valu
esar
ein
pp
m.
982
GARCIA et al. MAGMATIC PROCESSES FOR A PROLONGED KILAUEA ERUPTION
does not correlate with major and trace element con-centrations or ratios, or with the Pb, Sr or Nd isotopecompositions of Pu’u ’O’o lavas. Therefore, Pu’u ’O’omagmas probably partially assimilated and exchangedoxygen with a high-temperature metamorphosed Kilaueabasalt (Garcia et al., 1998a). The presence of basalticxenoliths with black olivines in some Pu’u ’O’o lavasmay be related to this process. The amount of crustalassimilation was greatest (perhaps up to 12%) during theearly period of magma mixing with rift-zone storedmagmas but decreased dramatically or stopped followingthe shift to continuous effusion in 1986 (Garcia et al.,1998a).
Magma mixing: episode 54 lavasPrevious studies of Kilauea’s historical eruptions havefocused on the products of short-lived eruptions andfound that magma mixing involving a differentiated, riftzone-stored magma and the influx of a more MgO-richmagma is a common process (e.g. Wright & Fiske, 1971).The lavas from the first 2 years of the Pu’u ’O’o eruptiondisplay ample petrographic, mineral and whole-rockchemical evidence of magma mixing (Garcia et al., 1992).Except for the rare high-Fo (87–89%) olivines, whichmay be relics from the influx of new magma, there is noevidence of magma mixing in Pu’u ’O’o lavas eruptedfrom episodes 30 to 54 (a 12 year period).
Episode 54 marked an abrupt change in lava min-eralogy and composition when evolved lavas with rareplagioclase phenocrysts (Tables 2 and 4) erupted a fewkilometers up-rift from the Pu’u ’O’o cone (Fig. 1). Thestart of episode 54 was accompanied by rapid subsidenceof Kilauea’s summit and an intrusion into the rift zone(Heliker et al., 1998a). The episode 54 lavas have distinctdifferences in mineralogy and whole-rock geochemistryfrom early 1983, evolved lavas (Figs 3–5), which wereerupted from the same area. These differences indicateFig. 5. Trace element variation in Pu’u ’O’o lavas based on ICP-MS
data (fields and symbols are as in Fig. 4; data from Table 6). Plots of that they were derived from compositionally distincthighly incompatible elements (e.g. La vs Ba) define linear trends with magmas. The mixing event for episode 54 lavas probablydecreasing values with time (except for episode 54 lavas). The scatter
occurred shortly before eruption. Otherwise, the high Foon this plot is almost within analytical error (2� error bars given forreference). There is an overall decrease in La/Yb ratio with time content olivines in the hybrid lavas would have been(except for episode 54 samples). The large range in Ba and La for the partially or completely re-equilibrated by diffusion. Noepisode 54 rocks is probably due to extensive crystal fractionation. The suitable mixing end members are evident among theplot of highly incompatible element ratios shows a tight data cluster
Pu’u ’O’o lavas for the episode 54 lavas because the Ffor episode 1–48 lavas but distinctly lower ratios for some episode50–55 lavas. vent sample does not lie on a mixing trend with the
other episode 54 lavas (Fig. 3). These results suggest thatKilauea’s east rift zone, a site of frequent intrusions overthe last 50 years (Klein et al., 1987), contains many closelythe matrix (Garcia et al., 1998a). This disequilibrium isspaced pockets of compositionally distinct magma thatthought to have occurred after the growth of olivinecan be forced to erupt by intruding dikes as suggestedbecause some of the olivines have oxygen isotope valuesby Ho & Garcia (1988).consistent with growth from a ‘normal’ mantle-derived
In summary, the dominant crustal processes modifyingmagma (>5·5‰; Garcia et al., 1998a). The size of theoxygen isotope disequilibrium between olivine and matrix the geochemistry of Pu’u ’O’o magmas are fractionation
983
JOURNAL OF PETROLOGY VOLUME 41 NUMBER 7 JULY 2000
Table 8: Pb, Sr and Nd isotopic data for selected Pu’u ’O’o eruption lavas
Sample Episode 206Pb/204Pb 207Pb/204Pb 208Pb/204Pb 87Sr/86Sr 143Nd/144Nd
29-Dec-92∗ 51 18·397 15·460 38·016 0·703601 0·512956
25-Apr-94∗ 53 18·407 15·472 38·058 0·703605 0·512957
27-Apr-95 53 18·427 15·492 38·105 0·703607 0·512955
10-Jan-97 53 18·414 15·480 38·067 0·703604 0·512953
10-Jan-98 55 18·429 15·502 38·139 0·703610 0·512950
All isotopic measurements were made using the VG Sector solid source mass spectrometer at the University of Hawaii. Ndand Sr isotopic fractionation corrections are 148Nd/144Nd = 0·241572 and 86Sr/88Sr = 0·1194. The data are reported relativeto standard values for La Jolla Nd (143Nd/144Nd = 0·511844) and NBS 981 Sr (87Sr/86Sr = 0·710258). Pb isotopic ratios arecorrected for fractionation using the NBS 981 Pb standard values of Todt et al. (1984). The estimated uncertainties (1�) arebased on repeated measurements of the standards: 206Pb/204Pb and 207Pb/204Pb ± 0·006, 208Pb/204Pb ± 0·017; 87Sr/86Sr ±0·000012, 143Nd/144Nd± 0·000009. Total procedural blanks are negligible: 10–80 pg for Pb, <120 pg for Sr and <20 pg for Nd.∗Data from Garcia et al. (1996).
Fig. 7. 206Pb/204Pb and 87Sr/86Sr ratios for recent Hawaiian tholeiiticlavas from Kilauea, Loihi and Mauna Loa volcanoes [data fromPietruszka & Garcia (1999a) for Kilauea, Garcia et al. (1993, 1995,1998b) for Loihi, and Kurz et al. (1995) and Rhodes & Hart (1995) forMauna Loa]. The wide scatter in Pb and Sr isotopes for recent lavasfrom these closely space volcanoes (triangle, Mauna Loa 1975 eruption;square, Loihi, 1996) requires at least three source components in theHawaiian plume. The Pu’u ’O’o lavas deviate with the well-definednegative trend for Mauna Loa and other Kilauea lavas; instead, theytrend towards and overlap with the Loihi field.
MANTLE-RELATED TEMPORALFig. 6. Variation in 206Pb/204Pb and 87Sr/86Sr with time for the Pu’u’O’o eruption (data from Table 7). Pb isotopes show the greatest GEOCHEMICAL VARIATIONSvariation compared with analytical error, with most of the variation
The whole-rock compositions of the non-hybrid Pu’uwithin the first 2 years of the eruption (a period of magma mixing).The Sr isotope ratio data define a similar but less well-defined trend ’O’o lavas from 1985 to 1998 were normalized to thebecause of the small variation in this ratio relative to analytical error same ‘parental’ MgO content to allow an evaluation to(see 2� error bars).
be made of the mantle-related geochemical variationsfor this eruption. A value of 10 wt % MgO was selectedfor normalization because it corresponds to the mostMgO-rich Pu’u ’O’o lava composition. Furthermore, theand accumulation of olivine. Crustal assimilation andhigher forsteritic olivines (85%) from this eruption are inmagma mixing have been important only during the
early part of the eruption and during episode 54. equilibrium with a melt of this MgO content, assuming
984
GARCIA et al. MAGMATIC PROCESSES FOR A PROLONGED KILAUEA ERUPTION
90% of the total iron is Fe2+. Because olivine is theonly phenocryst in these lavas, the normalization wasperformed using a mixture of 98·5% equilibrium olivineand 1·5% Cr-spinel, a typical proportion of these mineralsin Kilauea lavas (Wright, 1971), added in small in-crements (0·02%) to the liquid.
The results of this normalization procedure indicatethat highly incompatible elements (e.g. K) and CaOdecrease systematically with time (Fig. 8), Al2O3, Y andYb remain essentially constant, whereas Fe2O3 and SiO2
slightly increase. These results extend the trends thatwere identified for the 1985–1992 period of the Pu’u’O’o eruption (Garcia et al., 1996) and clearly demonstratea long-term systematic compositional variation for theeruption. A systematic decrease in the abundance ofincompatible elements was also noted for the1969–1974 Mauna Ulu eruption of Kilauea (Hofmannet al., 1984).
MANTLE MELTING MODELS FORTWO PROLONGED KILAUEAERUPTIONSThe magmas for the Pu’u ’O’o and Mauna Ulu riftzone eruptions of Kilauea are thought to have partiallybypassed the volcano’s summit magma storage reservoir(Ryan et al., 1981; Garcia et al., 1996). Thus, theseeruptions potentially offer a more direct view of theprocesses that generate tholeiitic basalts in the Hawaiianmantle plume compared with Kilauea’s historical summitlavas, which are thought to have been stored in thesummit reservoir for >30–120 years (Pietruszka & Gar-cia, 1999b). Unlike these summit lavas (Pietruszka &Garcia, 1999a), lavas from these two rift zone eruptionsdisplay a rapid decrease in the abundance of highlyincompatible elements (e.g. K, Nb) and in ratios of highlyto moderately incompatible trace elements (i.e. Ce/Yb)with little or no change in Pb, Sr and Nd isotope ratios(Hofmann et al., 1984; Figs 6 and 8). There are, however,important geochemical differences between these twoeruptions. The CaO contents of the Pu’u ’O’o eruption Fig. 8. Temporal variation of CaO and K2O contents (normalized tolavas decreased over time (Fig. 8), whereas no change in 10 wt % MgO, the most MgO-rich composition erupted from Pu’u
’O’o; see text for description of the normalization procedure), Ce/Ybthe CaO abundance of the Mauna Ulu lavas eruptedand Ba/Ce ratios for Pu’u ’O’o lavas [data from Tables 5 and 6 andfrom 1969 to 1971 was observed (Hofmann et al., 1984). Garcia et al. (1992, 1996)]. The overall decrease in these variables
Furthermore, ratios of highly incompatible trace elements during the course of the eruption should be noted. These changesrequire a depletion in the source for Pu’u ’O’o lavas with time. Χ (on(e.g. Ba/Ce) also decreased slightly over time for thethe ratio plots), samples used for the modeling shown in Table 8, whichPu’u ’O’o lavas, whereas these ratios are constant in thewas constrained to produce the observed ratios for Ce/Yb and Ba/
Mauna Ulu lavas (Hofmann et al., 1984; Fig. 8). The Ce. The 2� errors are given in the upper right corner of each plot.incompatible trace element ratio changes for both erup-tions are consistent with the relative incompatibility ofthese elements during mantle melting (e.g. Hauri et al., the compositional variation for these lavas. Below, we1994). This suggests that partial melting processes, rather evaluate two models that have been proposed to explain
the large compositional variation in prolonged Kilaueathan source heterogeneity, are the dominant controls on
985
JOURNAL OF PETROLOGY VOLUME 41 NUMBER 7 JULY 2000
eruptions with little or no change in Pb, Sr and Nd (Hofmann et al., 1984) and a decrease with time wasobserved for the Pu’u ’O’o eruption (Fig. 8).isotopes: batch melting of a homogeneous source (Hof-
mann et al., 1984) and progressive source depletionthrough mantle melting (Garcia et al., 1996).
Both melting models require an estimate of the degreeProgressive melting and source depletionof partial melting for the earliest Pu’u ’O’o lavas. WeAlternately, the temporal decrease in the incompatiblechose an initial melt fraction of 10% for Pu’u ’O’otrace element ratios (e.g. Ce/Yb and Ba/Ce) of Pu’ulavas based on our previous results of modeling Kilauea’O’o lavas may result from an increase in the proportionhistorical lavas (Pietruszka & Garcia, 1999a). This es-of melt derived from a depleted source component. Thetimate is derived from the systematic temporal variationsrelatively constant Pb, Sr and Nd isotope ratios of Pu’uin Pb, Sr and Nd isotope and incompatible trace element’O’o lavas suggest that this depletion is a recent con-ratios at this volcano over the last 200 years, which aresequence of melting within the Hawaiian plume. In thisthought to result from the short-term changes in thecontext, Garcia et al. (1996) proposed that the geochemicalcomposition of the mantle source and the degree ofvariations of Pu’u ’O’o lavas were caused by an eruption-partial melting. In the context of this model, Pu’u ’O’orelated modification of the Kilauea’s mantle sourcelavas formed at relatively high melt fractions comparedthrough progressive mantle melting and depletion. Theirwith other historical Kilauea lavas, on the basis of theirmodel assumed that the later Pu’u ’O’o lavas formed bylow ratios of highly over moderately incompatible tracea simple remelting of the same source that produced theelements (e.g. Ce/Yb). The source mineralogy and theearly lavas (e.g. at a constant melt fraction of 10%melting mode for the melting model are essentially thethroughout the eruption). In testing this model, however,same as used by Hofmann et al. (1984) for the Maunawe discovered that it fails to explain the overall chemistryUlu eruption. We found that the modeling results wereof Pu’u ’O’o lavas because the magmas generated fromrather insensitive to the melting mode, which was alsoa remelted source would have incompatible trace elementpointed out by Hofmann et al. (1984).abundances and ratios much lower than observed.
To overcome this problem, we considered a morecomplex progressive melting model using two source
Batch melting of a homogeneous source components (an ‘initial’ and a ‘depleted’) with the sameisotopic composition. The ‘depleted’ source componentThe temporal decrease in ratios of highly over moderately
incompatible trace elements for Mauna Ulu lavas can is assumed to have experienced a relatively small amountof prior melt removal (compared with the total meltbe explained by a 20% relative increase in the degree of
batch partial melting during the eruption (Hofmann et fraction). The ‘depleted’ source was mixed with the‘initial’ source, and this mixed source was partially meltedal., 1984). This model is consistent with the relatively
constant ratios of Sr and Nd isotopes and of highly to form the subsequent Pu’u ’O’o lavas. To determinethe possible extent of previous melting for the ‘depleted’incompatible trace elements (e.g. Ba/Ce) in lavas from
this eruption. A similar model for Pu’u ’O’o lavas (see source component, we evaluated a range of values from0·1 to 10% (i.e. slightly to strongly depleted sources,Table 9 for model parameters) also requires an >20%
relative increase in the degree of partial melting to respectively). Good residuals are obtained for in-compatible element abundances and ratios using thisexplain the variation in ratios of highly over moderately
incompatible trace element (Ce/Yb; Fig. 8). However, a model only if this parameter is <10%. If the amount ofprevious melting is >3%, the total degree of partialsimple increase in the degree of partial melting cannot
account for the slight temporal decrease of highly in- melting during the eruption remains nearly constant at>10± 1% (Table 9). For smaller or larger amounts ofcompatible trace element ratios observed for Pu’u ’O’o
lavas (e.g. Ba/Ce; Fig. 8) as a result of the relatively high recent melt removal, the model predicts that the degreeof partial melting during the Pu’u ’O’o eruption increasesmelt fraction expected for Kilauea tholeiites (5–10%;
Pietruszka & Garcia, 1999a). Furthermore, this batch or decreases over time, respectively. The maximumamount of recent melt removal permitted by the modelmelting model is inconsistent with experimental results
(Kushiro, 1996) for partial melting of a fertile garnet is>7% because, at higher values, the total melt fractionfor the late Pu’u ’O’o lavas would be <5%, which islherzolite source at 3 GPa (the mantle source conditions
that are thought to be needed for generation of Kilauea outside the range expected for Hawaiian shield lavas(5–20%; Watson, 1993). In all cases, the model resultsmagmas; Hofmann et al., 1984), which show that CaO
should increase with increasing degrees of partial melting suggest that the relative amount of the ‘depleted’ sourcecomponent increased progressively during the eruption,up to >20% (when cpx disappears). This expected
increase in CaO content with increasing degree of partial although magnitude depends on the amount of recentmelt removal to form the ‘depleted’ source. For 3%melting was not observed for the Mauna Ulu eruption
986
GARCIA et al. MAGMATIC PROCESSES FOR A PROLONGED KILAUEA ERUPTION
Tab
le9
:R
epre
sent
ativ
ere
sults
ofth
em
ixed
sour
cepr
ogre
ssiv
em
elting
mod
elfo
rP
u’u
’O’o
lava
s
So
urc
esM
od
elre
sult
s
2�er
ror
Init
ial
Dep
lete
dK
E30
-362
16-M
ar-8
726
-Mar
-89
12-M
ay-9
129
-Dec
-92
09-O
ct-9
407
-Ju
n-9
610
-Jan
-98
(%)
8-Fe
b-8
5(%
Dif
f)(%
Dif
f)(%
Dif
f)(%
Dif
f)(%
Dif
f)(%
Dif
f)(%
Dif
f)
Rb
(6·6
)0·
700·
0017
8·3
8·2
(1·8
)8·
0(2
·3)
7·4
(0·4
)7·
5(4
·6)
7·5
(0·1
)7·
2(0
·1)
6·7
(−3·
4)
Ba
(2·2
)9·
630·
0231
114
114
(0·3
)11
1(0
·7)
103
(1·1
)10
4(1
·2)
104
(0·2
)99
(−0·
6)92
(0·1
)
Th
(4·0
)0·
077
0·00
460·
900·
90(−
4·0)
0·88
(−1·
1)0·
82(−
8·2)
0·83
(−3·
9)0·
83(−
2·0)
0·80
(−4·
4)0·
74(−
0·7)
U(3
·4)
0·02
790·
0026
30·
322
0·32
1(−
0·3)
0·31
6(2
·1)
0·29
7(−
2·4)
0·30
0(2
·2)
0·29
9(−
0·2)
0·28
9(6
·0)
0·26
8(0
·8)
Nb
(2·4
)1·
170·
211
13·2
13·2
(3·9
)13
·0(1
·9)
12·4
(0·6
)12
·5(−
2·4)
12·5
(−4·
8)12
·1(−
2·9)
11·2
(−3·
9)
La(3
·4)
1·10
0·23
312
·312
·3(2
·2)
12·2
(0·5
)11
·7(0
·2)
11·7
(0·2
)11
·7(0
·0)
11·4
(0·0
)10
·6(1
·2)
Ce
(3·6
)2·
790·
803
30·3
30·4
(0·3
)30
·3(0
·7)
29·2
(1·1
)29
·3(1
·2)
29·4
(0·2
)28
·6(−
0·6)
26·5
(0·1
)
Pr
(3·4
)0·
434
0·18
94·
364·
39(0
·0)
4·43
(−1·
0)4·
35(−
0·6)
4·34
(−0·
3)4·
36(0
·9)
4·29
(0·3
)3·
96(0
·0)
Nd
(2·8
)2·
131·
1519
·820
·0(−
2·0)
20·3
(−0·
9)20
·2(−
0·3)
20·0
(0·0
)20
·2(0
·4)
19·9
(0·1
)18
·4(1
·3)
Sm
(2·2
)0·
662
0·46
05·
175·
25(−
1·0)
5·39
(−2·
4)5·
44(−
1·0)
5·35
(0·1
)5·
39(0
·9)
5·37
(0·8
)4·
97(1
·2)
Eu
(3·8
)0·
248
0·18
31·
801·
83(−
1·8)
1·88
(−1·
4)1·
91(2
·7)
1·87
(−0·
4)1·
89(−
0·4)
1·88
(−1·
1)1·
74(0
·0)
Gd
(2·8
)0·
863
0·68
15·
555·
65(0
·4)
5·84
(0·9
)5·
94(2
·6)
5·79
(−2·
8)5·
84(1
·0)
5·85
(3·3
)5·
43(0
·1)
Dy
(3·2
)1·
000·
868
4·94
5·04
(−0·
1)5·
24(−
0·8)
5·35
(1·1
)5·
18(−
0·1)
5·23
(2·7
)5·
25(−
0·2)
4·88
(1·6
)
Er
(4·0
)0·
658
0·60
52·
422·
48(−
0·7)
2·58
(−2·
5)2·
65(0
·7)
2·54
(−1·
4)2·
57(0
·7)
2·58
(−0·
9)2·
41(1
·2)
Yb
(2·6
)0·
687
0·65
11·
972·
02(0
·4)
2·11
(0·7
)2·
16(1
·1)
2·07
(1·2
)2·
09(0
·2)
2·11
(−0·
6)1·
97(0
·1)
�R
2(1
86)
(49·
0)(3
3·6)
(94·
2)(6
1·2)
(37·
5)(7
9·0)
(36·
4)
Ce/
Yb
4·07
1·23
15·4
15·1
14·4
13·5
14·2
14·1
13·6
13·5
Ba/
Ce
3·45
0·02
883·
773·
743·
663·
533·
563·
533·
473·
49
%To
tal
mel
t10
·010
·010
·09·
89·
39·
19·
29·
5
%D
eple
ted
sou
rce
0·0
2·7
9·4
20·2
19·6
21·8
25·3
23·5
Sam
ple
KE
30-3
62re
pre
sen
tsth
eea
rlie
stP
u’u
’O’o
lava
that
avo
ided
mag
ma
mix
ing
inth
eri
ftzo
ne
(Gar
cia
etal
.,19
96)
and
isth
ou
gh
tto
bes
tre
pre
sen
tth
ein
itia
lm
antl
e-d
eriv
edm
agm
aco
mp
osi
tio
nfo
rth
iser
up
tio
n.
Th
etr
ace
elem
ent
abu
nd
ance
so
fth
issa
mp
lew
ere
adju
sted
to16
wt
%M
gO
usi
ng
the
no
rmal
izat
ion
pro
ced
ure
des
crib
edin
the
text
tocr
eate
ap
ote
nti
alp
aren
tal
mag
ma
com
po
siti
on
.T
he
‘init
ial’
sou
rce
com
po
siti
on
was
calc
ula
ted
fro
mth
isp
aren
tal
mag
ma
assu
min
gth
atit
was
pro
du
ced
by
10%
no
n-m
od
alb
atch
par
tial
mel
tin
g(S
haw
,19
70)
of
aso
urc
eco
nta
inin
go
livin
e(6
0%),
clin
op
yro
xen
e(1
5%),
ort
ho
pyr
oxe
ne
(15%
)an
dg
arn
et(1
0%),
wh
ich
ente
rth
em
elt
in3:
4:2:
1p
rop
ort
ion
s.T
he
min
eral
og
y(6
1%o
livin
e,14
%cl
ino
pyr
oxe
ne,
15%
ort
ho
pyr
oxe
ne,
and
10%
gar
net
)an
dtr
ace
elem
ent
com
po
siti
on
of
the
‘dep
lete
d’
sou
rce
was
calc
ula
ted
fro
mth
is‘in
itia
l’so
urc
eas
sum
ing
that
3%m
elt
had
bee
nre
mo
ved
.T
hes
etw
oso
urc
eco
mp
on
ents
wer
em
ixed
(on
lyth
e%
of
the
dep
lete
dco
mp
on
ent
isre
po
rted
)an
dp
arti
ally
mel
ted
toa
deg
ree
(%to
talm
elt)
nec
essa
ryto
rep
rod
uce
the
ob
serv
edC
e/Y
ban
dB
a/C
era
tio
so
fth
ela
vas.
Th
ese
rati
os
wer
ese
lect
edb
ecau
seth
eyw
ere
det
erm
ined
wit
hre
lati
vely
hig
hp
reci
sio
nan
dar
ese
nsi
tive
toch
ang
esin
the
deg
ree
of
par
tial
mel
tin
gan
dso
urc
eco
mp
osi
tio
n,
resp
ecti
vely
.Fi
nal
ly,
the
calc
ula
ted
pri
mar
ym
agm
aco
mp
osi
tio
ns
wer
eev
olv
edb
yo
livin
eco
ntr
ol
tom
atch
the
ob
serv
edtr
ace
elem
ent
con
cen
trat
ion
so
fth
ela
vas.
Th
em
iner
al–l
iqu
idp
arti
tio
nco
effi
cien
tsu
sed
for
the
calc
ula
tio
ns
are
fro
mS
him
izu
&K
ush
iro
(197
5),
Hau
riet
al.(
1994
),Jo
hn
son
(199
8)an
dS
alte
rs&
Lon
gh
i(19
99)
for
gar
net
and
clin
op
yro
xen
e.T
he
Dva
lues
for
oliv
ine
and
ort
ho
pyr
oxe
ne
are
assu
med
tob
eze
ro.
For
each
lava
,th
em
od
elre
sult
sfo
rth
ein
com
pat
ible
trac
eel
emen
tco
nce
ntr
atio
ns
(in
pp
m),
and
resi
du
als
(%D
iff:
100×
[(ca
lcu
late
d/o
bse
rved
)−
1])
are
sho
wn
.T
he
�R
2va
lue
for
each
sam
ple
isth
esu
mo
fth
esq
uar
ed%
resi
du
als
for
all
elem
ents
.T
he
2�er
ror
for
each
trac
eel
emen
tis
bas
edo
nre
plic
ate
anal
yses
of
the
rock
stan
dar
d,
Kil1
919,
wh
ich
was
run
asan
un
kno
wn
ove
rth
eco
urs
eo
fth
isst
ud
y(s
eeP
ietr
usz
ka&
Gar
cia,
1999
a).
987
JOURNAL OF PETROLOGY VOLUME 41 NUMBER 7 JULY 2000
recent melt removal, which gave the lowest residuals, the systematic decrease in ratios of highly incompatible ele-ments cannot be explained by batch melting of a homo-amount of the ‘depleted’ source increases from 0 to
>25% (Table 9). Although the assumptions used in this geneous source, as was invoked for the Mauna Ulueruption, nor can a simple progressive melting modelmodeling are based on previous studies, they may seem
somewhat arbitrary. None the less, the modeling results explain them. Instead, Pu’u ’O’o magmas appear to havebeen produced by melting a mixed source with bothdemonstrate that the progressive melting model provides
a better explanation for the geochemical variation of the components having the same isotopic composition. Thedepleted source component was probably previously mel-Pu’u ’O’o lavas than other models we considered.
We favor >3% recent melt removal to form the ted by >3% relative to the initial source component.The percentage of the depleted component in Pu’u ’O’o‘depleted’ source component (Table 9) because (1) the
overall lava output rate has not changed during the Pu’u lavas progressively increasing during the eruption from0 to >25%. The depleted source must have formed’O’o eruption and (2) a small value for this factor is
required to account for the subtle temporal decrease in recently to have avoided a change in its Pb, Sr and Ndisotope composition but is probably not related to athe Th/U, Ba/Th, and Ba/U ratios of Pu’u ’O’o lavas
(as determined by high-precision isotope dilution Kilauea shield stage eruption because its extent of meltingwas too low. Given the isotopic similarity of lavas frommethods; A. Pietruszka, unpublished data, 1999). Phys-
ically, this removed melt could have been incorporated Pu’u ’O’o to many from the adjacent, younger volcano,Loihi, the melting region for the two volcanoes mayinto the early eruptive products of Pu’u ’O’o, a previous
Kilauea eruption, or, possibly, an eruption from the partially overlap.Prolonged eruptions such as Pu’u ’O’o offer an op-adjacent younger volcano, Loihi. If the amount of melt
removal was >3%, then it is unlikely to be related to a portunity for a better understanding of crustal and mantlemagmatic processes within active volcanoes. It will beKilauea eruption during its shield stage (last 400 ky;
Quane et al., 2000) because the range in melt fractions interesting to see if the lessons learned from this eruptionare applicable to eruptions at other oceanic island vol-for Hawaiian tholeiitic basalts is expected to be higher
(5–20%; Watson, 1993). Kilauea’s melting region may canoes.have progressively encroached upon a source that wasmelted to form Loihi’s alkalic and transitional lavas,which are thought to have formed by lower degrees of
ACKNOWLEDGEMENTSmelting than tholeiitic lavas (e.g. Watson, 1993). ThreeThis paper is dedicated to the memory of Keith Cox,additional lines of evidence favor a Loihi source: (1) Pu’uan icon of modern petrology. His friendship to students’O’o lavas are isotopically similar to some Loihi lavasof petrology was much appreciated. Our study would(Fig. 7); (2) U-series model results suggest that Pu’u ’O’onot have been possible without the assistance of manyis tapping a relatively large mantle source region, whichindividuals who collected samples (especially Frank Trus-may overlap with that of Loihi (Pietruszka et al., 2000);dell, Tom Hulsebosch, Scott Rowland and Marc Nor-(3) the mantle conduit for Kilauea dips south towardsman), and prepared and ran them for geochemicalLoihi Volcano (Tilling & Dvorak, 1993).analysis (P. Dawson, M. Vollinger, M. Chapman, B.Martin, J. Rayray, J. Parker, R. Magu and K. Kolysko-Rose). We would like to thank Friederike Klinge forassisting with microprobe analyses of episode 53 lavas,CONCLUSIONSK. Putirka for help in using his thermobarometer, andThe Pu’u ’O’o eruption is noteworthy in the historicalthe staff at the Hawaiian Volcano Observatory for theireruption record of Kilauea for its long duration, largediligent efforts in monitoring the Pu’u ’O’o eruption andvolume, and substantial lava geochemical variation.their cooperation. Constructive reviews by W. Bohrson,These features, and the simple mineralogy and pristineG. Fitton, A. Klugel and M. Wilson are gratefully ac-nature of the Pu’u ’O’o lavas, have allowed us to examineknowledged. This work was supported by NSF grants toboth crustal magmatic and mantle melting processes.M.G. (EAR-9315750 and EAR-9614247). This is SOESTThe dominant crustal processes influencing com-Contribution 5073.positional variations in Pu’u ’O’o lavas are olivine frac-
tionation and accumulation. Shallow magma mixing andcrustal assimilation were important only during the earlypart of the eruption and episode 54. The short-term REFERENCES(days to weeks) geochemical variations caused by olivine
Fodor, R. V., Keil, K. & Bunch, T. E. (1975). Contributions to thefractionation and accumulation are superimposed on mineral chemistry of Hawaiian rocks. IV. Pyroxenes in rocks fromlonger-term variations that are probably related to partial Haleakala and West Maui volcanoes. Contributions to Mineralogy and
Petrology 50, 173–195.melting processes in the mantle plume. The small but
988
GARCIA et al. MAGMATIC PROCESSES FOR A PROLONGED KILAUEA ERUPTION
Garcia, M. O. & Wolfe, E. W. (1988). Petrology of the erupted lava. Hofmann, A. W., Feigenson, M. D. & Raczek, I. (1984). Case studiesUS Geological Survey Professional Paper 1463, 127–143. on the origin of basalt, III. Petrogenesis of the Mauna Ulu eruption,
Garcia, M. O., Ho, R. A., Rhodes, J. M. & Wolfe, E. W. (1989). Kilauea 1969–1971. Contributions to Mineralogy and Petrology 88, 24–35.Petrologic constraints on rift zone processes: results from episode 1 Holcomb, R. (1987). Eruptive history and long-term behavior of Kilaueaof the Pu’u ’O’o eruption of Kilauea volcano, Hawaii. Bulletin of Volcano. US Geological Survey Professional Paper 1350, 261–350.Volcanology 52, 81–96. Johnson, K. T. M. (1998). Experimental determination of partition
Garcia, M. O., Rhodes, J. M., Ho, R. A., Ulrich, G. & Wolfe, E. W. coefficients for rare earth and high-field strength elements between(1992). Petrology of lavas from episodes 2–47 of the Pu’u ’O’o clinopyroxene, garnet, and basaltic melt at high pressures. Con-eruption of Kilauea Volcano, Hawaii: evaluation of magmatic pro- tributions to Mineralogy and Petrology 133, 60–68.cesses. Bulletin of Volcanology 55, 1–16. Kauahikaua, J., Mangan, M., Heliker, C. & Mattox, T. (1996). A
Garcia, M. O., Jorgenson, B., Mahoney, J., Ito, E. & Irving, T. (1993). quantitative look at the demise of a basaltic vent: the death ofTemporal geochemical evolution of the Loihi summit lavas: results Kupaianaha, Kilauea Volcano, Hawaii. Bulletin of Volcanology 57,from ALVIN submersible dives. Journal of Geophysical Research 98, 641–648.537–550. King, A., Waggoner, G. & Garcia, M. O. (1993). Geochemistry and
Garcia, M. O., Foss, D., West, H. & Mahoney, J. (1995). Geochemical petrology of basalts from ODP Leg 136. Ocean Drilling Program Scientificand isotopic evolution of Loihi Volcano, Hawaii. Journal of Petrology Results, 136. College Station, TX: Ocean Drilling Program, pp.36, 1647–1674. 107–118.
Garcia, M. O., Rhodes, J. M., Trusdell, F. A. & Pietruszka, A. P. Klein, F. W., Koyanagi, R. Y., Nakata, J. S. & Tanigawa, W. R.(1996). Petrology of lavas from the Pu’u ’O’o eruption of Kilauea (1987). The seismicity of Kilauea’s magma system. US GeologicalVolcano: III. The Kupaianaha episode (1986–1992). Bulletin of Survey Professional Paper 1350, 1019–1185.Volcanology 58, 359–379. Koyanagi, R. Y., Tanigawa, W. R. & Nakata, J. S. (1988). Seismicity
Garcia, M. O., Ito, E., Eiler, J. & Pietruszka, A. P. (1998a). Crystal associated with the eruption. US Geological Survey Professional Papercontamination of Kilauea Volcano magmas revealed by oxygen 1463, 183–235.isotope analysis of glass and olivine from the Pu’u ’O’o eruption Kurz, M. D., Kenna, T. C., Kammer, D., Rhodes, J. M. & Garcia,lavas. Journal of Petrology 39, 803–817. M. O. (1995). Isotopic evolution of Mauna Loa volcano: a view from
Garcia, M. O., Rubin, K. H., Norman, M. D., Rhodes, J. M., the submarine southwest rift. American Geophysical Union Monograph 92,Graham, D. W., Muenow, D. & Spencer, K. (1998b). Petrology and
289–306.geochronology of basalt breccia from the 1996 earthquake swarm
Kushiro, I. (1996). Partial melting of a fertile mantle peridotite at highof Loihi Seamount, Hawaii: magmatic history of its 1996 eruption.
pressures: an experimental study using aggregates of diamond.Bulletin of Volcanology 59, 577–592.
American Geophysical Union Monograph 95, 109–122.Ghiorso, M. S. & Sack, R. O. (1995). Chemical mass transfer in
Lofgren, G. (1980). Experimental studies on the dynamic crystallizationmagmatic systems IV. A revised and internally consistent thermo-
of silicate melts. In: Hargraves, R. B. (ed.) Physics of Magmatic Processes.dynamic model for the interpolation and extrapolation of liquid–solidPrinceton, NJ: Princeton University Press, pp. 487–551.equilibria in magmatic systems at elevated temperatures and pres-
Macdonald, G. A. (1944). Unusual features in ejected blocks at Kilaueasures. Contributions to Mineralogy and Petrology 119, 197–212.Volcano. American Journal of Science 242, 322–326.Gillard, D., Rubin, A. & Okubo, P. (1996). Highly concentrated
Macdonald, G. A. & Eaton, J. P. (1964). Hawaiian volcanoes duringseismicity caused by deformation of Kilauea’s deep magma system.1955. US Geological Survey Bulletin 1171, 1–85.Nature 384, 343–346.
Macdonald, G. A., Abbott, A. T. & Peterson, F. L. (1983). Volcanoes inHarris, A. J. L., Keszthelyi, L., Flynn, L. P., Mouginis-Mark, P. J.,the Sea. Honolulu, HI: University of Hawaii Press, pp. 89–81.Thornber, C., Kauahikaua, J., Sherrod, D., Trusdell, F., Sawyer,
Mangan, M. T., Heliker, C. C., Mattox, T. N., Kauahikaua, J. P. &M. W. & Flament, P. (1997). Chronology of the episode 54 eruptionHelz, R. T. (1995). Episode 49 of the Pu’u ’O’o–Kupaianahaat Kilauea Volcano, Hawaii, from GOES-9 satellite data. Geophysicaleruption of Kilauea volcano—breakdown of a steady-state eruptiveResearch Letters 24, 3281–3284.era. Bulletin of Volcanology 57, 127–135.Hauri, E. H., Wagner, T. P. & Grove, T. L. (1994). Experimental and
Moore, J. G. & Ault, W. U. (1965). Historic littoral cones in Hawaii.natural partitioning of Th, U, Pb and other trace elements betweenPacific Science 19, 3–11.garnet, clinopyroxene and basaltic melts. Chemical Geology 117, 149–
Pietruszka, A. P. & Garcia, M. O. (1999a). A rapid fluctuation in the166.mantle source and melting history of Kilauea Volcano inferred fromHeliker, C. C., Kauahikaua, J. P., Sherrod, D. R. & Thornber, C. R.the geochemistry of its historical summit lavas (1790–1982). Journal(1998a). The rise and fall of the Pu’u ’O’o cone, Kilauea Volcano,of Petrology 40, 1321–1342.1983–1998. EOS Transactions, American Geophysical Union 79, F956.
Pietruszka, A. P. & Garcia, M. O. (1999b). The size and shape ofHeliker, C., Mangan, M. T., Mattox, T. N., Kauahikaua, J. P. & Helz,Kilauea Volcano’s summit magma storage reservoir: a geochemicalR. T. (1998b). The character of long-term eruptions: inferences fromprobe. Earth and Planetary Science Letters 167, 311–320.episodes 50–53 of the Pu’u ’O’o–Kupaianaha eruption of Kilauea
Pietruszka, A. P., Rubin, K. H. & Garcia, M. O. (2000). 226Ra–Volcano. Bulletin of Volcanology 59, 381–393.230Th–238U disequilibria of historical Kilauea lavas (1790–1982) andHelz, R. T. & Thornber, C. R. (1987). Geothermometry of Kilaueathe dynamics of mantle melting within the Hawaiian plume. EarthIki lava lake, Hawaii. Bulletin of Volcanology 49, 651–668.and Planetary Science Letters (submitted).Ho, R. A. & Garcia, M. O. (1988). Origin of differentiated lavas at
Powers, H. (1955). Composition and origin of basaltic magma of theKilauea Volcano, Hawaii: implications from the 1955 eruption.Hawaiian Islands. Geochimica et Cosmochimica Acta 7, 77–107.Bulletin of Volcanology 50, 35–46.
Putirka, K. (1997). Magma transport at Hawaii: inferences based onHoffmann, J. P., Ulrich, G. E. & Garcia, M. O. (1990). Horizontaligneous thermobarometry. Geology 27, 69–72.ground deformation patterns and magma storage during the Pu’u
Putirka, K., Johnson, M., Kinzler, R., Longhi, J. & Walker, D. (1996).’O’o eruption of Kilauea volcano, Hawaii: episodes 22–42. Bulletin
of Volcanology 52, 522–531. Thermobarometry of mafic igneous rocks based on clinopyroxene–
989
JOURNAL OF PETROLOGY VOLUME 41 NUMBER 7 JULY 2000
liquid equilibria, 0–30 kbar. Contributions to Mineralogy and Petrology Tilling, R. I. & Dvorak, J. J. (1993). Anatomy of a basaltic volcano.Nature 363, 125–133.123, 92–108.
Quane, S. L., Garcia, M. O., Guillou, H. & Hulsebosch, T. P. (2000). Todt, W., Cliff, R. A., Hanser, A. & Hofmann, A. W. (1984). 202Pband 205Pb double spike for lead isotopic analyses. Terra Cognita 4,Magmatic history of the east rift zone of Kilauea Volcano, Hawaii
based on drill core from SOH 1. Journal of Volcanology and Geothermal 209.Ulmer, P. (1989). The dependence of the Fe2+–Mg cation-partitioningResearch (in press).
Rhodes, J. M. (1996). The geochemical stratigraphy of lava flows between olivine and basaltic liquid on pressure, temperatureand composition. Contributions to Mineralogy and Petrology 101,sampled by the Hawaii scientific drilling project. Journal of Geophysical
Research 101, 11729–11780. 261–273.Ulrich, G. E., Wolfe, E. W., Heliker, C. C. & Neal, C. A. (1987). Pu’uRhodes, J. M. & Hart, S. (1995). Episodic trace element and isotopic
variations in historical Mauna Loa lavas: implications for magma and ’O’o IV: evolution of a plumbing system. Hawaii Symposium 1987
Abstracts Volume, p. 259.plume dynamics. American Geophysical Union Monograph 92, 263–288.Roeder, P. L. & Emslie, R. F. (1970). Olivine–liquid equilibrium. Watson, S. (1993). Rare earth element inversion and percolation models
for Hawaii. Journal of Petrology 34, 763–783.Contributions to Mineralogy and Petrology 29, 275–289.Ryan, M. P., Koyanagi, R. Y. & Fiske, R. S. (1981). Modeling West, H. B., Gerlach, D. C., Leeman, W. P. & Garcia, M. O. (1987).
Isotopic constraints on the origin of Hawaiian lavas from the Mauithe three-dimensional structure of macroscopic magma transportsystems. Journal of Geophysical Research 86, 7111–7129. Volcanic complex, Hawaii. Nature 330, 216–220.
Wolfe, E. W., Garcia, M. O., Jackson, D. B., Koyanagi, R. Y., Neal,Salters, V. J. M. & Longhi, J. (1999). Trace element partitioning duringthe initial stages of melting beneath mid-ocean ridges. Earth and C. A. & Okamura, A. T. (1987). The Pu’u ’O’o eruption of Kilauea
volcano, episodes 1 through 20, January 3, 1983, to June 8, 1984.Planetary Science Letters 166, 15–30.Shaw, D. M. (1970). Trace element fractionation during anatexis. US Geological Survey Professional Paper 1350, 471–508.
Wolfe, E. W., Neal, C. A., Banks, N. G. & Duggan, T. J. (1988). GeologicGeochimica et Cosmochimica Acta 34, 237–243.Shimizu, N. & Kushiro, I. (1975). The partitioning of rare earth observations and chronology of eruptive events. US Geological Survey
Professional Paper 1463, 1–97.elements between garnet and liquid at high pressures: preliminaryexperiments. Geophysical Research Letters 2, 413–416. Wright, T. L. (1971). Chemistry of Kilauea and Mauna Loa lavas
in space and time. US Geological Survey Professional Paper 735, 1–45.Staudigel, H., Zindler, A., Hart, S. R., Leslie, T., Chen, C.-Y. &Clague, D. A., (1984). The isotope systematics of a juvenile intraplate Wright, T. L. & Fiske, R. S. (1971). Origin of the differentiated and
hybrid lavas of Kilauea Volcano, Hawaii. Journal of Petrology 12,volcano: Pb, Nd, and Sr isotope ratios of basalts from Loihi seamount,Hawaii. Earth and Planetary Science Letters 69, 13–29. 1–65.
990